Skip to main content

Okurigana: When Kanji Bleeds Into Hiragana

Okurigana are the hiragana written immediately after a kanji to complete a word. They most often attach to the kun'yomi reading of the kanji stem.123 They are why べる is not just : the kanji carries the meaning, and the kana tail tells you which Japanese word the kanji is standing in for.14

Overview

What okurigana is in one paragraph

A standard dictionary gloss reads: "Kana written after a kanji to complete the full (usually kun) reading of the word; declensional kana ending."3 The single kanji しょく by itself cannot tell the reader whether to read taberu (to eat), kuu (to eat, plain), or shoku (a meal, Sino-Japanese). The hiragana tail べる identifies the word as べる.14

The term itself is morphologically transparent, meaning its parts show how it works. おく仮名がな literally means "sent-along kana": おくり is the continuative form of the verb おくる (okuru, "to send, dispatch, append"), and 仮名かな is voiced to がな by sequential voicing (rendaku) inside the compound.56

べる7
"to eat." The kanji 食 carries the meaning; the kana tail べる carries the inflection.

たか8
"tall" or "expensive." The kanji 高 holds the lexical sense; the い inflects, becoming かっ in past たかかった and く in adverbial たかく.

Okurigana, furigana, and kana: the terms sorted

Three terms share the word "kana," which can confuse learners at first. Okurigana sits on the baseline as part of the word; the kanji and the kana together spell one lexeme, or dictionary word.12 Furigana sits above the kanji in horizontal writing, or to the right in vertical writing, as a reading aid outside the word itself.1

Both are written in hiragana most of the time, but they do different jobs. Okurigana is morphological: it is part of how the word is spelled. Furigana is annotational: it tells the reader how to pronounce a kanji whose reading they might not know.14

Why okurigana exists: one kanji, several spoken words

A single jōyō kanji (a kanji on Japan's standard-use list) typically carries one or more kun'yomi and one or more on'yomi. Without okurigana, a bare kanji in running text would be ambiguous between several distinct Japanese words.14

Take the kanji うえ. It is shared by noboru (のぼる, "to climb"), agaru (がる, "to rise"), ageru (げる, "to raise, give"), ue (うえ, "top"), and the on'yomi compound 上手じょうず (jōzu). The kana tail picks which spoken word the kanji stands for in this position.14

Wikipedia and the sci.lang.japan FAQ describe the function in similar terms: okurigana "inflect adjectives and verbs, and force a particular kanji to have a specific meaning and be read a certain way."12

のぼ14
"to climb / ascend (a slope, stairs, a rank)." Tail る on 上 selects the noboru reading.

がる14
"to rise / go up (intransitive)." Tail がる on the same 上 selects the agaru reading.

げる14
"to raise / give (transitive)." Tail げる on 上 selects the ageru reading; the kanji alone is unreadable as a verb.

Form: where the kana tail starts

Verbs: the tail starts at the inflecting mora

The principal rule from 通則 1 (General Rule 1) is short: 「活用かつようのあるは、活用かつよう語尾ごびおくる」 ("words that conjugate take okurigana on the inflectional ending").910 The kana run begins no later than the mora that changes when the verb conjugates.

For an ichidan (ru-verb) like べる, the inflecting region is everything from べ onward. The kanji 食 stays in place while the kana tail changes for each form.78

FormSpellingReading
Dictionary食べるたべる
Past食べたたべた
Negative食べないたべない
Te-form食べてたべて
Polite食べますたべます

For a godan (u-verb) like く, the inflecting region is just く. A single kana mora carries the entire conjugation.78

FormSpellingReading
Dictionary行くいく
Past行ったいった
Negative行かないいかない
Polite行きますいきます

あさはんべます。7
"I eat breakfast."

学校がっこうく。78
"I go to school."

あさはやきる。78
"I get up early in the morning."

I-adjectives: the tail is at least the final い

The same 通則 1 applies to i-adjectives. The inflecting kana (い in the dictionary form, く in the adverbial, かっ in the past) is okurigana, and the kana run starts no later than the inflecting mora.910

たかい splits as たか + い. The kanji 高 carries the lexical sense; the い inflects across the paradigm: たかい / たかくない / たかかった / たかくて.8

あたらしい splits as あたら + しい. The tail covers everything from the inflecting syllable し onward. The kanji 新 on its own is the on'yomi morpheme, or meaningful word part, shin used in compounds like 新聞しんぶん. The kun'yomi adjective requires しい to be readable as atarashii.98

このほんたかい。8
"This book is expensive."

あたらしいくるまった。8
"I bought a new car."

試験しけんむずかしかった。8
"The exam was difficult."

Na-adjectives and nouns: usually no tail

The principal rule from 通則 3 is equally short: 「名詞めいしは、おく仮名がなけない」 ("nouns do not take okurigana").910 Words that do not inflect generally take no tail: 元気げんき, しずか, やま, かわ.91

The exception is a small class of nouns derived from inflecting words. These keep the source word's okurigana under 通則 4: うごき (from うごく), たり (from たる), ねがい (from ねがう).910

元気げんき子供こども8
"a healthy child." No okurigana on the na-adjective stem 元気.

しずかな部屋へや8
"a quiet room." The か is conventionally part of the stem and is written as kana by historical practice, but no further tail is added.

やま8
"to look at the mountain." Bare-kanji noun, no okurigana.

Okurigana lives on kun'yomi, not on'yomi

Cross-source consensus is unusually clean here: okurigana attaches almost exclusively to kun'yomi (native Japanese) readings.124 Sino-Japanese morphemes in on'yomi compounds do not inflect in Japanese and therefore do not need a tail. Wikipedia states the rule directly: "okurigana are used only with kun'yomi (native Japanese readings), not with on'yomi (Chinese readings)."1

The contrast is the cleanest single mnemonic a learner can carry: べる (kun, taberu) takes べる, but 食事しょくじ (on + on, shokuji) takes nothing. く (kun, iku) takes く, but 行進こうしん (on + on, kōshin) takes nothing. あらわす (kun, arawasu) takes す, but 表現ひょうげん (on + on, hyōgen) takes nothing.14

The kun/on test is a heuristic, not an absolute

通則 2 covers a small set of compounds where a derived word built on a kun'yomi base preserves its interior okurigana inside a larger compound (, ながぼし).910 The "no tail on on'yomi" rule handles the overwhelming majority of cases at sight. Treat the kun-derived compound exceptions as a second pass once the main rule is automatic.

毎日まいにちあさはんべる。7
"I eat breakfast every day."

食事しょくじ時間じかんです。8
"It's mealtime." No tail on 食事 because the reading is on'yomi.

The MEXT (Ministry of Education, Culture, Sports, Science and Technology) okurigana rules at a glance

The 1973 Cabinet notification and its 1981 and 2010 amendments

The current standard is 「おく仮名がなかた」, issued as 内閣告示第二号 on June 18, 1973 (昭和48年).11121310 It replaced an earlier 「おくりがなのつけかた」 issued as 内閣告示第一号 in 1959 (昭和34年). The 1959 standard was superseded entirely, not merely amended.1415

A 1981 (昭和56年) amendment was issued in coordination with the first 常用じょうよう漢字かんじひょう (Jōyō Kanji Table) notification of the same year.11161 The most recent amendment is 平成22年内閣告示第3号, issued November 30, 2010 (平成22年), aligned with the revised 常用漢字表 of the same date.1116

Public-document use sits under a parallel 平成22年内閣訓令第1号, "公用文こうようぶんにおける漢字かんじ使用しようとうについて." This order constrains official writing to the principal forms (本則) of 通則 1 through 6 and the conventional forms (通則 7). The 許容 (permitted alternatives) are allowed in official documents only for non-conjugating compound words that cannot be misread.161718

The seven 通則 (general rules), pitched for learners

The notification sets seven general rules (七つの通則) for okurigana assignment. Each has a 本則 (principal form) and, where needed, 例外 (exception) and 許容 (permitted alternative) clauses. Together, the rules list about 500 sample words.910

Each rule, in plain terms:

  • 通則 1, 「活用かつようのあるは、活用かつよう語尾ごびおくる」: conjugating words (verbs, i-adjectives, and the small set of inflecting na-adjective-like words) take okurigana on the inflecting region. Example: く, きる, たかい, しずかだ.910
  • 通則 2 covers derived conjugating words: they preserve the okurigana of the contained source word. Example: うごかす from うごく.910
  • 通則 3, 「名詞めいしは、おく仮名がなけない」: nouns take no okurigana. Example: つき, とり, はな, やま. The 例外 (exceptions) include うしろ, さいわい, なかば, みずから, わざわい.910
  • 通則 4 handles nouns derived from inflecting words: they keep the source word's okurigana. Example: うごき, たり, ねがい, はやさ.910
  • 通則 5 covers adverbs, adnominals (noun-modifying words), and conjunctions: the kana tail is on the final mora. Example: かならず, きたる (adnominal kitaru, "next, coming"), および.910
  • 通則 6 covers compound words: they inherit each constituent's okurigana as if written alone. Example: く, もうむ, わせる.91910
  • 通則 7 covers a defined set of compound nouns whose okurigana-less form is conventionally fixed for signage, forms, and technical or legal use. Example: 取扱とりあつかい, 申込もうしこみ, 受付うけつけ, 引換ひきかえ, 切手きって, 葉書はがき.91910

The two 付表 rules for jōyō-list special readings

The 付表ふひょう (attached table) handles fixed-form readings that fall outside the seven 通則 because they are listed as whole-word readings in the 常用漢字表 付表. Textbook examples include 明日あす (asu), 今日きょう (kyō), 大人おとな (otona), 田舎いなか (inaka), and 二十歳はたち (hatachi).111

Public-document use treats these "付表の語" as a closed list governed by the principal forms with no okurigana variation, alongside 通則 1 through 6 本則 and 通則 7.1617 For a learner, the point is recognition rather than memorization: when 今日 reads as kyō with no tail, the form is not missing okurigana. It is following a separate rule list.

What the rules are and are not for

The Bunkachō explanation states the notification's own scope: it is a 「よりどころ」 (basis, reference standard) for okurigana assignment in 「一般いっぱん社会しゃかい生活せいかつにおいて、現代げんだい国語こくごあらわすための」 writing (writing modern Japanese in ordinary public life), not a binding prescription for all writing.119

The preface explicitly excludes 「科学かがく技術ぎじゅつ芸術げいじゅつその各種かくしゅ専門せんもん分野ぶんや個々人ここじん表記ひょうき」 (scientific, technical, artistic, and other specialized fields, and individual personal writing).119

This scope note is what licenses the 許容 variants. Traditional, literary, and specialist writing legitimately uses forms (such as おこなう, あらわす) that the 本則 would not pick first.1715

Common ambiguity and permitted-variant cases

行う vs 行なう

Both forms are recognized by the 1973 Cabinet notification. おこなう is the 本則 under 通則 1: the kana run begins at the inflecting mora う. おこなう is the 許容 (permitted alternative). It places the kana run one mora earlier so the reader sees the inflection clearly, even in conjugated forms like おこなった rather than おこった.141510

A historical reversal is worth knowing. In the 1959 notification (内閣告示第一号, 昭和34年), the principal form was 行なう and 行う was the permitted alternative. The 1973 revision flipped the hierarchy.1415

Modern textbook practice follows the 本則: 光村みつむら図書としょ and the major school publishers print 行う uniformly.14 Government and legal practice is stricter. 内閣ないかく法制局ほうせいきょく法令ほうれいにおける漢字かんじ使用しようとうについて」 (2010) does not adopt the 許容, so legal text uses only 行う.1718

Use 行なう sparingly outside specialist contexts

行なう is fully grammatical as a 通則 1 「許容」, but school textbooks and the 内閣法制局 standard for legal text use only the 本則 行う. If you write 行なう in a school essay or a business document, it will read as deliberately conservative or as a holdover from the 1959 standard, even though no rule is broken. Default to 行う; recognize 行なう when you meet it in older government print or traditional writing.171415

会議かいぎおこなう。14
"We hold a meeting." 本則 form.

会議かいぎおこなう。1510
"We hold a meeting." 許容 form; same word, longer tail.

表す vs 表わす

あらわす is the 本則 under 通則 1; あらわす is the 許容 listed in the 1973 notification.1510 Both write the same word, arawasu. The longer form makes the inflection visible at the cost of an extra kana.15

The same paired pattern appears in あらわす / あらわす, あらわれる / あらわれる, ことわる / ことわる, and たまわる / たまわる.1510

感謝かんしゃ気持きもちをあらわす。15
"I express my feelings of gratitude." 本則 form.

感謝かんしゃ気持きもちをあらわす。15
"I express my feelings of gratitude." 許容 form; identical reading and meaning.

著しい vs 著るしい and the broader "permitted variant" family

The 1973 notification's 通則 1 「許容」 clause lists いちじるしい as 本則 alongside the permitted longer いちじるしい for the adjective ichijirushii (remarkable, striking).91510

The pattern is general, not lexical. Any verb or i-adjective whose principal form leaves the inflection at the very end may have a permitted variant that pushes the kana run one mora earlier, if the Bunkachō word list licenses that particular pair. That licensing limits the family: it is not every conjugating word, only the specific entries enumerated in the notification.1510

Three pairs (行う / 行なう, 表す / 表わす, 著しい / 著るしい) are representative; do not extrapolate a longer-form rule to verbs the notification does not list.

いちじるしい変化へんかられる。1510
"A remarkable change can be seen." 本則 form.

いちじるしい変化へんかられる。1510
"A remarkable change can be seen." 許容 form.

Compound-word omissions: 取扱 vs 取り扱い, 申込 vs 申し込み

通則 7 licenses a closed set of compound nouns to drop interior okurigana entirely: 取扱, 申込, 受付, 引換, 切手, 葉書, 場合, 立場, 場所.91910 The same word in running prose generally keeps the okurigana. Examples include あつかい (the noun in a sentence), あつかう (the verb), もうみ (the noun in a sentence), and もうむ (the verb).1910

The split is by register, not by region. The same word can appear with or without interior okurigana on the same page, depending on whether it is a heading, a sign, a form label, or a flowing sentence.

RegisterFormNotes
Sign, form label, fixed compound取扱通則 7, conventional spelling
Running prose, noun取り扱い通則 6 本則
Running prose, verb取り扱う通則 6 本則
Sign, form label申込通則 7
Running prose, noun申し込み通則 6 本則
Running prose, verb申し込む通則 6 本則

The 平成22年内閣訓令第1号 narrows the public-document scope further. 通則 7 applies as the principal rule for the listed compounds. 通則 6's 許容 (dropping interior okurigana on a non-conjugating compound where there is no misreading risk) is allowed where reading is unambiguous.1617

取扱とりあつかい注意ちゅうい17
"Handle with care." Fixed signage compound; no interior okurigana per 通則 7.

この書類しょるいあつかいに注意ちゅういしてください。19
"Please be careful in handling this document."

申込書もうしこみしょ提出ていしゅつする。19
"I submit the application form." Fixed compound 申込 (per 通則 7) plus the noun-forming 書.

Reading-disambiguating tails: 上る vs 上がる, 下る vs 下がる, 細い vs 細かい

The kana tail can do more than signal conjugation: it can pick out which of two distinct words a kanji represents. のぼる (noboru) vs がる (agaru), くだる (kudaru) vs がる (sagaru), ほそい (hosoi) vs こまかい (komakai), and がる (magaru) vs げる (mageru) all share a kanji and a related meaning field. But they are distinct verbs or adjectives with their own conjugation paradigms.14

This is the strongest single argument for why okurigana is mandatory in the modern system. Without the tail, written Japanese could not distinguish noboru from agaru on the page. The kanji 上 alone is silent on which word the writer means.

Always write 上がる, not 上る, for the agaru reading

For the noboru / agaru pair specifically, あがる must be written 上がる. The intransitive form is forced into the longer spelling so that its kanji-tail logic stays parallel to the transitive げる; writing it as 上る would collide with the noboru reading and lose the distinction the rule exists to create.4

階段かいだんのぼる。14
"I climb the stairs." Tail る selects noboru.

階段かいだんがる。14
"I go up the stairs." Tail がる selects agaru.

ほそいと4
"a thin thread." Tail い selects hosoi.

こまかい4
"small / fine print." Tail かい selects komakai; the same kanji 細 reads differently.

Good to know

"Bleeds into hiragana" is a feature, not a transcription compromise

The hiragana tail is doing morphological work the kanji cannot. Kanji are logographic and do not encode conjugational endings on their own. Okurigana is the mechanism by which a Sinographic stem joins the Japanese verbal and adjectival inflection system. Treating okurigana as "the part you didn't bother to write in kanji" misframes the system and slows reading, because every tail carries part-of-speech and inflection information the kanji is structurally unable to express.12

When in doubt, copy the dictionary headword

The principal forms (本則) in the 1973 notification are the forms used as headwords by major monolingual dictionaries (大辞泉だいじせん, 広辞苑こうじえん) and by Jisho.153 Permitted variants (許容) exist in print, but they are not the form to learn first; school textbooks teach only the 本則.14 If your default is "spell it the way the dictionary headword spells it," you will align with both school practice and the public-document standard with no extra effort.

IME typing and okurigana

Typing the full reading on a Japanese IME (input method editor), for example taberu, surfaces the inflecting form 食べる as the first conversion. Typing only the stem, for example tabe, surfaces 食べ as a noun-stem candidate (used in compounds like もの).1

The okurigana is what makes the IME's first guess match the part of speech you wanted. Skipping it on input is one of the easier ways to end up with the wrong word on screen.

Furigana sits above, okurigana sits beside

A reader arriving from the topic of furigana should expect okurigana to behave as part of the word's spelling and to be present even when furigana is absent.1 Furigana is annotation outside the line of text; okurigana is part of the word on the line.14 The two never compete for the same role.

Dropping okurigana on a conjugating verb

If you write 食る for taberu, you have dropped 通則 1's requirement that the inflecting region of a conjugating word be in kana. The kanji 食 alone is unreadable as a verb because it holds the stem; the inflection lives in the tail. The correct form is the dictionary spelling:

べる92
"to eat" (dictionary form).

The same principle catches the parallel error on i-adjectives (たか for takai) and every other conjugating word.

Writing a fixed signage compound with interior okurigana

On a sign labeled "Reception," writing け instead of 受付 reads as the wrong register, not as a different word. 通則 7 lists 受付 as a fixed conventional noun with no interior okurigana for signage, forms, and headlines.1910

The same word in running prose ("窓口まどぐちけています") keeps the okurigana per 通則 6. The split is by register, not by meaning.

Why the word 送り仮名 means "sent-along kana"

おくり is the continuative form of おくる (okuru, "to send, append, dispatch"); 仮名かな voices to がな by rendaku inside the compound.56 The metaphor is that the kana is sent along after the kanji to finish the word. The term's parts hint at its function.

Kun'yomi takes a tail; on'yomi does not

A single one-line test handles most beginner okurigana decisions: if the kanji is read in its native Japanese (kun'yomi) sense, expect a kana tail; if the kanji sits in a Sino-Japanese compound (on'yomi), expect no tail. べる vs 食事しょくじ; く vs 行進こうしん; あらわす vs 表現ひょうげん.14 The exceptions live in 通則 2's kun-derived compound list. They matter on a second pass, not first.

What this article does not cover

Rendaku, ateji, gikun, full verb and adjective conjugation paradigms, and the 常用漢字表 itself are related topics that each warrant their own article. This piece confines itself to the okurigana standard and its rule set. The larger writing-system map sits in the wider Japanese-learning index of how kanji, hiragana, and katakana share a sentence.91

See also

References

Footnotes

  1. Wikipedia. "Okurigana." https://en.wikipedia.org/wiki/Okurigana 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

  2. sci.lang.japan FAQ. "What is okurigana?" https://www.sljfaq.org/afaq/okurigana.html 2 3 4 5 6

  3. Jisho.org entry for 送り仮名: "Kana written after a kanji to complete the full (usually kun) reading of the word; declensional kana ending." https://jisho.org/word/%E9%80%81%E3%82%8A%E4%BB%AE%E5%90%8D 2 3

  4. Japanese with Anime. "okurigana 送り仮名." (limitation: language-learner blog; used only for examples cross-checked against Bunkachō text.) https://www.japanesewithanime.com/2017/12/okurigana.html 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

  5. Wiktionary. Entry for 送り仮名. Etymology section (continuative form of 送る + 仮名, rendaku-voiced). https://en.wiktionary.org/wiki/%E9%80%81%E3%82%8A%E4%BB%AE%E5%90%8D 2

  6. Wiktionary. Entry for 送る (okuru), Japanese verb senses including the linguistics sense "to add okurigana." https://en.wiktionary.org/wiki/%E9%80%81%E3%82%8B 2

  7. Banno, Eri et al. Genki: An Integrated Course in Elementary Japanese, vol. 1, 3rd ed. The Japan Times, 2020. Verb-conjugation chapter (Lessons 3, 8, 9) for 食べる, 行く, ます-form examples. 2 3 4 5 6 7

  8. Makino, Seiichi and Tsutsui, Michio. A Dictionary of Basic Japanese Grammar. The Japan Times. Entries for 食べる (ichidan / ru-verb), 行く (godan / u-verb with irregular て-form), and the 高い adjective paradigm. 2 3 4 5 6 7 8 9 10 11 12 13 14

  9. 文化庁. 「送り仮名の付け方」解説. https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/okurikana/kaisetu.html 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

  10. wordrabbit. 「送り仮名の付け方・ルール」. https://wordrabbit.jp/blog/52 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

  11. 文化庁. 「送り仮名の付け方」内閣告示(昭和48年内閣告示第二号、昭和56年・平成22年一部改正). Index page. https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/okurikana/index.html 2 3 4 5 6

  12. 文化庁. 「送り仮名の付け方」訓令・告示制定文. https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/okurikana/kunrei.html

  13. 国立国会図書館. 当用漢字音訓表・送り仮名の付け方:内閣告示・内閣訓令(大蔵省印刷局、昭和48年). https://ndlsearch.ndl.go.jp/books/R100000002-I000001207006

  14. 光村図書出版. 「教科書の言葉Q&A 第19回 行う、それとも『行なう』?」. https://www.mitsumura-tosho.co.jp/webmaga/kotoba-to-manabi/kotoba/detail19 2 3 4 5 6 7

  15. 文化庁. 「送り仮名の付け方」本文 通則1(本則・例外・許容). Lists 行う/行なう, 表す/表わす, 現れる/現われる, 断る/断わる, 賜る/賜わる, 著しい/著るしい as 本則 / 許容 pairs under 通則 1. https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/okurikana/honbun01.html 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

  16. 文化庁. 「送り仮名の付け方」解説 (kaisetu01): 「常用漢字表」の内閣告示に伴う一部改正について. https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/okurikana/kaisetu01.html 2 3 4 5

  17. 参議院法制局. 「法令における送り仮名」. コラム065. https://houseikyoku.sangiin.go.jp/column/column065.htm 2 3 4 5 6 7

  18. 内閣法制局. 「法令における漢字使用等について」(平成22年内閣法制局総総第208号). Cited via 17. 2

  19. 文化庁. 「送り仮名の付け方」本文 通則6. https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/okurikana/honbun06.html 2 3 4 5 6 7