Researched and fact-checked through the J-Compass editorial process; our editorial team reviews sourcing and accuracy before publishing. See our methodology →
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 食べる is not just 食: 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 食 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 食べる.14
The term itself is morphologically transparent, meaning its parts show how it works. 送り仮名 literally means "sent-along kana": 送り is the continuative form of the verb 送る (okuru, "to send, dispatch, append"), and 仮名 is voiced to がな by sequential voicing (rendaku) inside the compound.56
食べる7
"to eat." The kanji 食 carries the meaning; the kana tail べる carries the inflection.
高い8
"tall" or "expensive." The kanji 高 holds the lexical sense; the い inflects, becoming かっ in past 高かった and く in adverbial 高く.
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 上. It is shared by noboru (上る, "to climb"), agaru (上がる, "to rise"), ageru (上げる, "to raise, give"), ue (上, "top"), and the on'yomi compound 上手 (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
上る14
"to climb / ascend (a slope, stairs, a rank)." Tail る on 上 selects the noboru reading.
上がる14
"to rise / go up (intransitive)." Tail がる on the same 上 selects the agaru reading.
上げる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: 「活用のある語は、活用語尾を送る」 ("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 食べる, the inflecting region is everything from べ onward. The kanji 食 stays in place while the kana tail changes for each form.78
| Form | Spelling | Reading |
|---|---|---|
| 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 行く, the inflecting region is just く. A single kana mora carries the entire conjugation.78
| Form | Spelling | Reading |
|---|---|---|
| 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 |
朝ご飯を食べます。7
"I eat breakfast."
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
高い splits as 高 + い. The kanji 高 carries the lexical sense; the い inflects across the paradigm: 高い / 高くない / 高かった / 高くて.8
新しい splits as 新 + 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. The kun'yomi adjective requires しい to be readable as atarashii.98
この本は高い。8
"This book is expensive."
新しい車を買った。8
"I bought a new car."
試験は難しかった。8
"The exam was difficult."
Na-adjectives and nouns: usually no tail
The principal rule from 通則 3 is equally short: 「名詞は、送り仮名を付けない」 ("nouns do not take okurigana").910 Words that do not inflect generally take no tail: 元気, 静か, 山, 川.91
The exception is a small class of nouns derived from inflecting words. These keep the source word's okurigana under 通則 4: 動き (from 動く), 当たり (from 当たる), 願い (from 願う).910
元気な子供8
"a healthy child." No okurigana on the na-adjective stem 元気.
静かな部屋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.
山を見る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: 食べる (kun, taberu) takes べる, but 食事 (on + on, shokuji) takes nothing. 行く (kun, iku) takes く, but 行進 (on + on, kōshin) takes nothing. 表す (kun, arawasu) takes す, but 表現 (on + on, hyōgen) takes nothing.14
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 (落ち葉, 流れ星).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.
毎日朝ご飯を食べる。7
"I eat breakfast every day."
食事の時間です。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 「送り仮名の付け方」, issued as 内閣告示第二号 on June 18, 1973 (昭和48年).11121310 It replaced an earlier 「送りがなのつけ方」 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 常用漢字表 (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, "公用文における漢字使用等について." 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, 「活用のある語は、活用語尾を送る」: conjugating words (verbs, i-adjectives, and the small set of inflecting na-adjective-like words) take okurigana on the inflecting region. Example: 書く, 生きる, 高い, 静かだ.910
- 通則 2 covers derived conjugating words: they preserve the okurigana of the contained source word. Example: 動かす from 動く.910
- 通則 3, 「名詞は、送り仮名を付けない」: nouns take no okurigana. Example: 月, 鳥, 花, 山. The 例外 (exceptions) include 後ろ, 幸い, 半ば, 自ら, 災い.910
- 通則 4 handles nouns derived from inflecting words: they keep the source word's okurigana. Example: 動き, 当たり, 願い, 速さ.910
- 通則 5 covers adverbs, adnominals (noun-modifying words), and conjunctions: the kana tail is on the final mora. Example: 必ず, 来る (adnominal kitaru, "next, coming"), 及び.910
- 通則 6 covers compound words: they inherit each constituent's okurigana as if written alone. Example: 書き抜く, 申し込む, 打ち合わせる.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, 申込, 受付, 引換, 切手, 葉書.91910
The two 付表 rules for jōyō-list special readings
The 付表 (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 (asu), 今日 (kyō), 大人 (otona), 田舎 (inaka), and 二十歳 (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 「一般の社会生活において、現代の国語を書き表すための」 writing (writing modern Japanese in ordinary public life), not a binding prescription for all writing.119
The preface explicitly excludes 「科学・技術・芸術その他各種専門分野や個々人の表記」 (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 行なう, 表す) 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. 行う is the 本則 under 通則 1: the kana run begins at the inflecting mora う. 行なう 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 行なった rather than 行った.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: 光村図書 and the major school publishers print 行う uniformly.14 Government and legal practice is stricter. 内閣法制局 「法令における漢字使用等について」 (2010) does not adopt the 許容, so legal text uses only 行う.1718
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
会議を行う。14
"We hold a meeting." 本則 form.
会議を行なう。1510
"We hold a meeting." 許容 form; same word, longer tail.
表す vs 表わす
表す is the 本則 under 通則 1; 表わす 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 著す / 著わす, 現れる / 現われる, 断る / 断わる, and 賜る / 賜わる.1510
感謝の気持ちを表す。15
"I express my feelings of gratitude." 本則 form.
感謝の気持ちを表わす。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 著しい as 本則 alongside the permitted longer 著るしい 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.
著るしい変化が見られる。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 取り扱い (the noun in a sentence), 取り扱う (the verb), 申し込み (the noun in a sentence), and 申し込む (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.
| Register | Form | Notes |
|---|---|---|
| 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
取扱注意17
"Handle with care." Fixed signage compound; no interior okurigana per 通則 7.
この書類の取り扱いに注意してください。19
"Please be careful in handling this document."
申込書を提出する。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. 上る (noboru) vs 上がる (agaru), 下る (kudaru) vs 下がる (sagaru), 細い (hosoi) vs 細かい (komakai), and 曲がる (magaru) vs 曲げる (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.
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 上げる; writing it as 上る would collide with the noboru reading and lose the distinction the rule exists to create.4
細い糸4
"a thin thread." Tail い selects hosoi.
細かい字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, 広辞苑) 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 食べ物).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:
The same principle catches the parallel error on i-adjectives (高 for takai) and every other conjugating word.
Writing a fixed signage compound with interior okurigana
On a sign labeled "Reception," writing 受け付け 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 ("窓口で受け付けています") 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"
送り is the continuative form of 送る (okuru, "to send, append, dispatch"); 仮名 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. 食べる vs 食事; 行く vs 行進; 表す vs 表現.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
- The Te-Form in Japanese: Construction Rules
- な-Adjective Conjugation in Japanese: All Tenses and Forms
- Jukugo (熟語): How Kanji Combine to Form Japanese Words
- How to Predict the Reading of an Unknown Kanji Compound: The On+On Default, Jūbako, Yutō, and the Look-It-Up Bucket
- Rendaku in Kanji Compounds: Why 紙 Becomes -gami
- Nanori (名乗り): The Name-Only Kanji Readings That Break the On/Kun Rules