The dictionary I'm using lists two common pronunciations of 「九」: 「きゅう」 and 「く」. I've usually heard it as 「きゅう」 before, but when saying 「9時」, it always seems to be 「くじ」. Is there a rule to this?


Today, きゅう is the default (i.e. productive) on-yomi pronunciation of 九 (or 9) for counting most things, and only a small portion of words requires く.

hours (o'clock) (9時, 19時), dates (19日, 29日), month name (9月)

Preferred or alternative to きゅう:
hours (duration) (9時間, 19時間...), years (9年(間), 2009年...), people (9人, 19人...), degree (29度, 39度...), bare double-figure numbers (19, 29, 1029...), -fold, -ple (9重, 19重...), followed by (第9次, 9次元...)

Sporadic traditional or euphonic examples (Google it for details):
九段, 九十九里, 四十九日, 九分九厘, 九条, 九九, 九曜, 第九, 月9 (= 月曜9時) etc.
(There are far more old words; historically く had been the prevailing pronunciation until it was replaced because of its homophony with 苦{く} "anguish".)

  • 2
    Regarding the last paragraph, while I find this explanation interesting, as きゅう is the 漢音 and く the 呉音, I wonder whether the homophony with 苦 really has been the main driving forces behind this change from く to きゅう, or whether this is mostly a case of folk etymology. Are you aware of any reasearch or references for this? – blutorange Dec 15 '14 at 16:53
  • 1
    In fact I tried to find some bases while was writing the answer, and I couldn't get any sources from the web. So it's not more than what I'm told from people for now. What I can say for certain is that きゅう is just peculiar one while other numerals under 兆 are in accord on having 呉音. – broccoli forest Dec 15 '14 at 17:46
  • 2
    @blutorange I came at an 1917 book that says 二{ふた}, 四{よん}, 七{なな}, 九{きゅう} was introduced "to avoid mishearing". kindai.ndl.go.jp/info:ndljp/pid/926955/38 (Of course, ふた in digit-reading is obsolete except in stock market or phonetic code.) – broccoli forest Jan 19 '15 at 1:22

Generally speaking, there is no hard rule to decide which reading of a kanji is used for a given word or compound. To be certain, you need to look it up in a dictionary and remember each word on a case-by-case basis.

However, there are certain tendencies that allow you to guess, better than by random guessing, which reading to use for a certain kanji. To put it another way, you can use your feeling for the language to come up with an educated guess for a new word, and need to remember only the exceptions, reducing the required mental load and capacities.

Another point I believe is worth mentioning, you need to learn each word anyway, be there kanji or not. By learning a word, I am referring to the association between meaning and sound. You need to do that for any other language as well. And once you know the word, all you need to do when reading a text with kanji is identifying the word in question, and you will know its pronunciation. For example, if you have learned the word kuji = 🕘, and you see 九時 in some written text, your knowledge that 九 = 9, 時 = time together with the context allows you to conclude that 九時 = 🕘. Therefore, even without remembering which reading to use for each word, you know that 九時 = kuji.

Tendencies include the following:

ON/KUN reading:

  • ON, in kanji compounds, eg. 芸者【げいしゃ】, 恍惚【こうこつ】
  • ON reading if it has become an individual word on its own*, eg. 肉【にく】, 秒【びょう】, 魔【ま】, 豹【ひょう】, 雹【ひょう】, 生【せい】, 僕【ぼく】, 本【ほん】, 印【いん】
  • KUN reading if there are okurigana forcing it, eg. 頂【いただ】きます, 行【おこな】う, 躊躇【ためら】う (cf. 躊躇【ちゅうちょ】), 話【はな】し
  • KUN for individual kanji whose ON reading is not an individual words, especially if it is a noun with the okurigana of the 連用形 omitted, eg. 神【かみ】, 鏡【かがみ】, 話【はなし】, 歌【うた】, 鉈【なた】
  • words consisting of multiple words etymologically are prone to mixed readings, especially when the ON reading exists as a word, or it contains pre/suffixes: 取調室【とりしらべしつ】, 秒読み【びょうよみ】, 消印【けしいん】, 団子【だんご】, 客間【きゃくま】, 時代物【じだいもの】
  • Words referring to cultural activities, rituals, objects, food, clothing, medicine, plants, weaponry etc. tend to use irregular readings, and you had better look them up, eg. 八咫烏【やたがらす】, 地謡座【じうたいざ】, 栴檀板【せんだんのいた】, 上四方固【かみしほうがため】, 天照大神【あまてらすおおみかみ】, 剪定鋏【せんていばさみ】

Some compound words without okurigana read KUN: 子守唄【こもりうた】, 神々【かみがみ】, 素直【すなお】, 取引【とりひき】. Note that there are some mixed ON/KUN compounds (cf. 湯桶読み, 重箱読み).

Different ON readings:

  • Most kanjis have got one ON reading that is much more common in contemporary Japanese. Often, this is the KAN (漢音) reading. Some of the more frequent kanjis have got more common readings, but the rarer, the less irregular, and most kanji are infrequent**. Eg., 植【しょく】, 衛【えい】, 療【りょう】, 恥【ち】, 攪【かく】, 命【めい】, 焉【えん】, 集【しゅう】. Words with other readings are not so common: 命終【みょうじゅう】, 駢植【へんち】, 兵衛【ひょうえ】, 結集【けちじゅう】, 攪拌【こうはん】
  • GO (呉音) reading for Buddhistic vocabulary, and common words of Buddhistic origin: 薬師来【やくしにょらい】, お経【きょう】, 槃【ねはん】, 脱【げだつ】,八道【はっしょうどう】, 進【しょうじん】, 名色【みょうしき】, 苦【おんしょうえく】, 不動王【ふどうみょうおう】
  • Some older (Heian/Kamakura - Edo) words use TOU/SOU (唐宋音), eg. 椅【いす】, 杏【あんず】, 灯【あんどん】, 箪【たんす】, 九時【くじ】

(*) Often, these words are used only as part of some larger sentence and might be unintelligible if used in isolation. For example, この世に生をうける is understandable, but saying せい without context is not.

(**) Some less frequent kanji with a reading different than what their structure (semantic-phonetic compound) suggests have acquired a new ON reading corresponding to the phonetic element. This is called 慣用音, eg. 消耗【しょうもう】 instead of the original 消耗【しょうこう】.

  • 1
    Yes, I'm familiar with how to guess what reading a kanji uses, and I know all about おん vs. くん readings, but part of my question was meant to ask what reading 九 uses when it's used on its own, as I've heard both of its おん readings used. – Rurfs Dec 14 '14 at 23:56
  • 漢 reading of 九 is きゅう, 呉 is く. On its own, 九 means 9 and is usually read きゅう (or ここの as in なな、や、ここの、とお, but you probably wouldn't usr kanji here), く occurs mostly only in some compounds, eg 十中八九. Dictionary く: 〔「きゅう」に比べて使用範囲が狭く、熟した表現において用いられることが多い〕 – blutorange Dec 15 '14 at 6:33

Is there a rule to this?

Yes! The rule is 9時 is always くじ. (And similarly, 9月 "September" is always くがつ. Cf. 9ヶ月 きゅうかげつ "nine months".)


A single Kanji usually has several different readings, and numbers 1-10 are probably some of the most salient examples of this. The readings can vary wildly. For example,

九{きゅう} (nine) 九本{きゅうほん} (nine bottles) 九番{きゅうばん} (problem number 9) 九時{くじ} (nine o'clock) 九月{くげつ} (September -- 9th month) 九日{ここのか} (the 9th day of the month) 九つ{ここの} (nine objects that are either round or of no discernible shape)

I can't really think of a general rule, unfortunately. I can say, however, you'll eventually get used to all the exceptions.

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.