Skip to main content

Long Vowels in Katakana: How the Chōonpu ー Works and Why Hiragana Doesn't Use It

Long vowels in katakana are written with a single mark, the chōonpu (長音符 ー), which holds the preceding vowel for one extra mora (timing beat).1 One symbol covers every vowel row. That is why コーヒー, ラーメン, and ケーキ look so tidy, and why hiragana does not use the same bar in standard prose.23

Overview​

What the chōonpu is​

The mark is called 長音符 (chōonpu), literally "long-sound symbol."14 Japanese sources also attest the alternate names 音引き (onbiki), 棒引き (bōbiki), 伸ばし棒 (nobashi-bō, "stretching stick"), and 長音記号 (chōonkigō).4 The Unicode Consortium designates it the "Katakana-Hiragana Prolonged Sound Mark."1

Two codepoints carry the glyph: U+30FC for the fullwidth ー and U+FF70 for the halfwidth ｰ.14 The fullwidth form is JIS X 0208 kuten 01-28. It is the form learners meet in print, online text, and IME output.1

The operational definition is short. ー "indicates a chōon, or a long vowel of two morae in length," meaning the preceding katakana vowel is held for one additional mora.14 It is the only spelling device modern katakana uses for long vowels, regardless of which vowel is being lengthened.21

The official name is one term, the everyday name another

長音符 (chōonpu) is the term in government documents and reference works. 伸ばし棒 (nobashi-bō, "stretching stick") is the term you will hear in a Japanese classroom.4 Both name the same mark. "Stretching stick" is also the more useful mental image for a beginner.

Where you meet it​

The chōonpu appears chiefly inside katakana-written items: gairaigo (外来語, loanwords), foreign personal and place names, onomatopoeia stylised in katakana, and brand names.214 These are the categories assigned to katakana, so they are also the categories that use the bar.

The 1991 Cabinet Notification 「外来語の表記」 makes the chōonpu the default device for long vowels in loanwords. The text reads "長音は，原則として長音符号「ー」を用いて書く" ("Long vowels are, as a rule, written using the long-vowel mark 'ー'"), and its example list includes エネルギー, オーバーコート, グループ, ゲーム, ショー, テーブル, パーティー.2

朝あさはコーヒーを飲のみます。5
"I drink coffee in the morning."

The mark also appears in hiragana, but only in informal contexts: interjections like あー, onomatopoeia like どすーん, dialect spellings like てめー, and emphatic lengthening like ながーい.4 Manga and graphic dialogue favour the bar in these uses because it shows held vowels visually. Standard hiragana prose instead follows the doubled-vowel rules of the 1986 Cabinet Notification 「現代仮名遣い」.34

The JLPT (Japanese-Language Proficiency Test) does not test the chōonpu by itself. Still, the N5 vocabulary list contains more than thirty katakana loanwords whose spelling depends on the bar (コーヒー, コート, パーティー, セーター, シャワー, スカート, スポーツ, テーブル, テープ, タクシー).6 Every katakana reading task at N5 assumes you know it.6

The rule in katakana​

How the mark works​

One mark for any vowel. The chōonpu extends whatever katakana vowel precedes it by exactly one mora.214 The same codepoint serves all five vowel rows; the reader infers the held vowel from the kana that sits to the left of the bar.1

The 1991 內閣告示 sets this as the working rule for loanwords, with examples spanning every vowel row: エネルギー, オーバーコート, グループ, ゲーム, ショー.2 The mark applies uniformly across base kana and the extended-katakana inventory of 第2表 (combinations such as ティ, ファ, ウィ that arose for foreign-sound transcription).27

The contrast with hiragana is structural, not stylistic. Hiragana spells long vowels by adding a vowel kana from the matching row (ああ, いい, うう). The o-row defaults to う (おう), while a closed exception class takes お (おお).3 Katakana drops the per-vowel pattern and puts the work into one mark.21

Worked examples across the five vowels​

Each vowel row has exactly one chōonpu form: the row's kana plus ー. The table below lists a beginner-level loanword in every row, drawn from the 1991 內閣告示 and the JLPT N5 vocabulary list.26

Vowel rowFormExampleReadingGloss
aカーカードkādocard
iキーコーヒーkōhīcoffee
uクースープsūpusoup
eケーケーキkēkicake
oコーコートkōtocoat

Several N5 words use more than one bar. スーパー (sūpā, "supermarket") has ー in two places, コーヒー (kōhī, "coffee") has one ー after コ and another after ヒ, and オーバーコート (ōbākōto, "overcoat") strings three together.26 Romanization in Modified Hepburn writes each chōonpu as a macron on the lengthened vowel.18

Extended-katakana clusters pair with ー the same way base kana do: ティー (tī, "tea"), パーティー (pātī, "party"), ウィークデー (wīkudē, "weekday").26 The chōonpu rule does not care which table the preceding kana comes from.27

Reading the beat​

A long vowel is one vowel phoneme held over two morae, not a stressed or louder syllable.89 Phonetic duration is roughly 2.5 to 3 times that of a short vowel.8 The ー itself counts as the second mora, the held continuation of whatever vowel precedes.18

So コーヒー is four morae, not two syllables. The picture below shows the beats.

Vowel length is phonemic, which means it can flip meaning by itself. The minimal pairs below all turn on a single chōonpu.

Short formReadingGlossLong formReadingGloss
ビルbirubuildingビールbīrubeer
ヒル (蛭)hiruleechヒールhīruheel

ビルとビールは違ちがいます。9
"Biru and bīru are different words."

Two morae, one vowel quality

ー is not a fresh vowel pulse. Reading コーヒー as ko-o-hi-i in four distinct vowel attacks is wrong. The correct reading is kōhī, one held o followed by one held i.8 Modified Hepburn writes both morae as a single macron-bearing vowel for exactly this reason.8

Why hiragana does not use ー​

Two scripts, two solutions​

Katakana marks length with one universal symbol; hiragana marks length with a per-row pattern. The split is governed by two separate Cabinet Notifications.23

The 1986 Cabinet Notification 「現代仮名遣い」 第1 5 gives no rule for using ー in hiragana. It treats hiragana long vowels strictly as a per-row pattern.3 The 1991 Cabinet Notification 「外来語の表記」 names ー as 長音符号 and assigns it as the default device for the script of loanwords, namely katakana.2 The two documents form a deliberate split.23

The two notations are not interchangeable in standard prose. A loanword written in katakana keeps the bar (コーヒー). The same form recast in hiragana for a children's book follows hiragana rules (こうひい / こおひい, not こーひー).314

Why the split exists​

The division of labour is functional. Katakana's modern job is to transcribe material from foreign sound systems where vowel length is supplied by the source language. One universal mark works across whatever vowel turns up.24 Hiragana spells native morphology where the underlying vowel quality matters historically.3

The o-row split between おう (default) and おお (closed exception list) preserves the etymology of words like 多い, 遠い, 氷, 通る. Their historical kana had a separate o-class kana (ほ, を) that has since merged in pronunciation but remains visible in spelling.3 Hiragana's per-row pattern is the residue of that history. Katakana, which carries no such native morphology, does not need to record it.

Late Middle Japanese monophthongisation, the merging of vowel sequences into single long vowels, affected /au/, /ou/, /eu/, and /iu/. It forced the 1946 「現代かなづかい」 reform to choose spellings for the merged sounds.10 The choice fell on doubled vowels and オ列 + う for hiragana, and on the chōonpu for katakana.310 The two scripts inherited two different solutions to the same phonological history.10

When hiragana borrows ー​

The Japanese Wikipedia entry on 長音符 lists the registers where hiragana ー is at home: 感動詞 (interjections like あー, ありゃー), 擬音・擬態語 (onomatopoeia and mimetics like どすーん, そーっ, あーん), 方言・俗語 (dialect and slang like てめー, あぶねーっ, あちー), and 語調の強調による長呼 (emphatic lengthening like ながーい, よーく, たかーい).4 Manga and graphic-narrative writing favour the bar for these uses because it shows held vowels in dialogue visually.4

やったー！4
"Yes! / I did it!"

ながーい道みち。4
"A loooong road."

Outside these registers, hiragana ー is not standard. Formal prose, textbooks, and news copy follow 「現代仮名遣い」 第1 5 and use doubled vowels.34

Hiragana ー belongs to informal registers only

やったー, すごーい, おはよー read as casual or expressive in dialogue and manga.4 The same content in formal prose follows the standard pattern: やった, すごい, おはよう.3 Mixing the two registers is the most common slip in learner writing.

Vertical-text orientation​

Horizontal vs. vertical forms​

The chōonpu uses the same Unicode codepoint (U+30FC) regardless of writing direction.14 The font and layout engine rotate the glyph 90 degrees when the text runs vertically: "日本語の縦書き表記の中で長音を使う際には，90度回転させた形で表記する" ("when long vowels are used in vertical Japanese writing, they are written in a form rotated 90 degrees").4

In horizontal text (横書き, yokogaki) the mark is a horizontal bar; in vertical text (縦書き, tategaki) it is a vertical bar.14 Historically, the vertical form preceded the horizontal: classical Japanese print was set vertically, and the bar was drawn vertically before yokogaki became common.4

What this means in practice​

Learners who have only read horizontal web text often meet the vertical form for the first time in novels, manga, signage, and tategaki magazine layouts.14 The orientation may surprise you, but it does not change identity, length, or sound. The mark is still U+30FC, still one mora long, and still bound to the preceding kana.14

新聞しんぶんでニュースを読よみます。6
"I read the news in the newspaper."

In a printed newspaper this same line appears in tategaki, with the ー of ニュース drawn vertically.14 The halfwidth ｰ (U+FF70) does not rotate the same way in vertical text and is rarely used in tategaki layouts. Fullwidth ー is the canonical form for both directions.1

Loanword exceptions: エイ vs エー​

The default: long English /eɪ/ becomes エー​

The 1991 內閣告示 makes the chōonpu the default for English long vowels and diphthongs in loanwords, including the [eɪ] family.211 The official examples include ゲーム (gēmu, "game"), テーブル (tēburu, "table"), テープ (tēpu, "tape"), ペーパー (pēpā, "paper"), and ケーキ (kēki, "cake").2

For English word-final unstressed /ə/-set (-er, -or, -ar), the official rule is given explicitly. The text reads "英語の語末の－er, －or, －ar などに当たるものは，原則としてア列の長音とし長音符号「ー」を用いて書き表す。ただし，慣用に応じて「ー」を省くことができる" ("Items corresponding to English word-final -er, -or, -ar are, in principle, written as a-row long vowels using the long-vowel mark 'ー'; the 'ー' may be omitted in accordance with established convention").211 Examples: エレベーター / エレベータ, コンピューター / コンピュータ.211

The exception: エイ for diphthong-like preservation​

The 1991 內閣告示 carves out an explicit exception clause: "「エー」「オー」と書かず，「エイ」「オウ」と書くような慣用のある場合は，それによる" ("When there is an established convention of writing 「エイ」「オウ」 rather than 「エー」「オー」, follow that convention").211

The diagram below shows the decision an editor makes when an English /eɪ/ word arrives.

Source-attested loanwords commonly spelled with エイ rather than エー include エイト (eito, "eight"), ペイント (peinto, "paint"), レイアウト (reiauto, "layout"), スペイン (Supein, "Spain"), ケインズ (Keinzu, "Keynes"), and サラダボウル (sarada bōru, "salad bowl"; note オウ for the same reason).21112

The pattern is not derived from a sound rule. The official text says explicitly that when publisher or editorial convention has settled on エイ/オウ, that convention wins over the default ー.211

Why you see both spellings of the same word​

Convention changes over time and differs by publisher, so the same word can appear in two accepted spellings.131112

メイク vs メーク is the textbook ambiguous case. The NHK 日本語発音アクセント新辞典 (NHK Publishing, 2016) records メーキャップ as the broadcast-standard form, continuing the entry from the 1998 edition.13 Outside NHK, both メイク and メーク are in use. Cosmetics-industry copy tends to prefer メイク (matching the エイ exception), while broadcast journalism leans on メーク.1311

コンピューター vs コンピュータ is the other model case, but it concerns trailing ー rather than エイ. The 1991 內閣告示 sanctions both, and JIS Z 8301 explicitly says "長音符号は，用いても略しても誤りでない" ("Using or omitting the long-vowel mark is not incorrect").11

Government and broadcast usage keep ー. Pre-2008 engineering style guides drop it.11 In 2008, METI (the Ministry of Economy, Trade and Industry) revised its 技術情報通則 so that the trailing ー is recommended for terms three morae or longer in general writing, aligning industrial style with national broadcast usage.11

The 文化審議会国語分科会 「外来語（カタカナ）表記ガイドライン 第3版」 (2015) reaffirms that the agency does not require a single form. Each field follows its own convention, and the underlying 1991 內閣告示 authorises this.212

明日あしたのパーティーのペイントを買かいました。26
"I bought paint for tomorrow's party."

Here パーティー takes the default ー, while ペイント follows the convention exception spelling. Both come from the same official document.2

A note on native words written in katakana​

When a native Japanese word is rendered in katakana for emphasis, in scientific binomial nomenclature (two-part scientific names), or in onomatopoeia, long vowels still take ー, not the hiragana doubled-vowel pattern.14 Examples include ラーメン (originally a Chinese loan, fully naturalised, written in katakana with ー), ハイカー (a loanword hybrid sometimes stylised this way), and コーラ (a katakana-only word despite some claims of native status).14

The script governs the spelling, not the etymology.14 Words conventionally written in hiragana follow hiragana rules even when their referents are foreign. Words conventionally written in katakana follow katakana rules even when their referents are native.314

Katakana onomatopoeia in manga and advertising use ー the same way loanwords do (どすーん, あーん).4 These are also the forms the Wikipedia 長音符 entry uses to illustrate the bar's informal hiragana use, since the same forms are sometimes printed in hiragana for stylistic effect.4

Good to know​

Think "stretch bar," not "dash"​

The alternate Japanese name 伸ばし棒 (nobashi-bō, "stretching stick") captures the function: ー is one mora of held vowel, not punctuation, not a hyphen, not a pause.4 The goal is to read コーヒー as four morae (コ・ー・ヒ・ー), not as two syllables with decoration.148

ー is not the ASCII hyphen, em dash, or kanji 一​

Several visually similar characters are entirely distinct in code and function: the ASCII hyphen-minus (-, U+002D), the em dash (U+2014), and the kanji 一 ("one", U+4E00). The chōonpu is U+30FC and sits at the kana baseline, next to the preceding kana.1 In IME input on a JIS keyboard, the dedicated long-vowel key produces ー directly. Typing ka-do yields カード.1 Writing パ-ティ- with ASCII hyphens is a common copy-paste artefact and reads as broken text. The correct form is パーティー.

Long vowels carry their own mora for pitch​

The chōonpu counts as a full mora in the pitch-accent domain.8 This is why コーヒー and a hypothetical コヒー would sound different to a Japanese ear before meaning is even considered: the long o is one extra beat of pitch territory. Beginners do not need to drill pitch on long vowels. Still, knowing the mora carries pitch explains why treating long vowels as "decoration" leads to consistently off-rhythm speech.8

Decoding ー as a separate vowel sound​

Reading コーヒー as ko-o-hi-i in four distinct vowel attacks misses the point of the mark. The correct reading is kōhī: one held o followed by one held i.89 The second mora is a continuation of the first vowel, not a new sound. That is why Modified Hepburn collapses both morae into one macron-bearing vowel.8

コーヒーはおいしいです。56
"Coffee is delicious."

メイク and メーク are both correct​

A learner who claims メイク is wrong because the default rule yields ー has the rule but not the exception. Both spellings are allowed by the 1991 內閣告示, with the エイ exception clause authorising メイク wherever publisher convention prefers it.211 NHK's broadcast-standard reference favours メーキャップ. Cosmetics-industry copy commonly prints メイク.13 Neither is a mistake. They are two accepted conventions.

Dropping the trailing ー (ユーザ, コンピュータ)​

Engineering style guides, especially JIS Z 8301 and the pre-2008 編集要綱 it informed, permit dropping the final ー from -er/-or/-ar loanwords. They cite the omission clause in the 1991 內閣告示.211 General-purpose Japanese (newspapers, broadcast, government documents, school textbooks) retains the ー: ユーザー, コンピューター, プリンター.11 In 2008 METI revised 技術情報通則 to recommend the trailing ー for terms three morae or longer in general writing, narrowing the gap between industrial and broadcast style.11

Bōbiki: when hiragana did use ー​

From 1900 (Meiji 33) to 1908 (Meiji 41), the Ministry of Education's 小学校令施行規則 mandated 棒引き仮名遣い (bōbiki kanazukai, "bar notation kana spelling") in elementary-school textbooks.1415 Under that system, long vowels in Sino-Japanese words were written with ー regardless of script: 学校 was spelled がっこー, 校長 こーちょー, in lines like ホントー デス カ.1415 The reform was abolished in 1908 and never restored. That is why hiragana ー survives only as the casual marker it remains in modern Japanese.1415

See also​

References​

Footnotes​

  1. Wikipedia. "Chōonpu." Names, Unicode codepoints, vertical orientation, hiragana use. https://en.wikipedia.org/wiki/Ch%C5%8Donpu ↩ ↩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. 文化庁. 「外来語の表記」 (平成3年6月28日内閣告示第2号 / 1991 Cabinet Notification No. 2). 留意事項その2 (細則的な事項), 注3 (長音). https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/gairai/honbun06.html ↩ ↩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 ↩28

  3. 文化庁. 「現代仮名遣い」 (昭和61年7月1日内閣告示第1号 / 1986 Cabinet Notification No. 1). 本文 第1 5(1)–(5) (long-vowel rules per vowel row). https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/gendaikana/honbun_dai1.html ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11 ↩12 ↩13

  4. ウィキペディア日本語版. 「長音符」. 別名 (音引き, 棒引き, 伸ばし棒, 長音記号), Unicode (U+30FC, U+FF70), 縦書き 90 度回転, 仮名遣いの歴史, 平仮名での俗用 (感動詞・擬音・漫画). https://ja.wikipedia.org/wiki/%E9%95%B7%E9%9F%B3%E7%AC%A6 ↩ ↩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 ↩28 ↩29 ↩30 ↩31 ↩32

  5. JLPT Sensei. "コーヒー (koohii), JLPT N5 vocabulary." https://jlptsensei.com/learn-japanese-vocabulary/%E3%82%B3%E3%83%BC%E3%83%92%E3%83%BC-koohii-meaning/ (limitation: aggregator, used only as N5 alignment and natural-sentence verification source) ↩ ↩2

  6. JLPT Sensei. "JLPT N5 Katakana Words List." Confirms コーヒー, コート, パーティー, セーター, シャワー, スカート, スポーツ, テーブル, テープ, タクシー at N5. https://jlptsensei.com/jlpt-n5-katakana-words-list/ (limitation: aggregator, used only for level alignment) ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8

  7. ウィキペディア日本語版. 「外来語の表記」. 1991 年告示の構成 (前書き・本文・付録), 第1表・第2表・留意事項. https://ja.wikipedia.org/wiki/%E5%A4%96%E6%9D%A5%E8%AA%9E%E3%81%AE%E8%A1%A8%E8%A8%98 ↩ ↩2

  8. Vance, Timothy J. The Sounds of Japanese. Cambridge University Press, 2008. Chapter 5 (Moras), pp. 56–66, on long vowels as two-mora units. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11

  9. Wikipedia. "Japanese phonology." Section on vowel length and the mora; ヒール (hīru "heel") vs 蛭 (hiru "leech") minimal pair. https://en.wikipedia.org/wiki/Japanese_phonology ↩ ↩2 ↩3

  10. Frellesvig, Bjarke. A History of the Japanese Language. Cambridge University Press, 2010. ISBN 9780521653206. Chapter 8 on Late Middle Japanese monophthongisation; discussion of the historical conditions that led modern hiragana to spell long o as オ列 + う rather than with a length mark. ↩ ↩2 ↩3

  11. 黒田祐我. 「Userはユーザかユーザーか: カタカナによる外来語の表記、特に長音表記について」. u-site (ユーザビリティ・サイト). Discussion of 1991 内閣告示第2号, JIS Z 8301, and the エイ/オウ exception clause. https://u-site.jp/lecture/notation-of-foreign-words ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11 ↩12 ↩13 ↩14 ↩15

  12. 文化審議会国語分科会. 「外来語（カタカナ）表記ガイドライン 第3版」. 2015 年. 制定: 2015年8月 / 発行: 2015年9月. https://www.bunka.go.jp/seisaku/bunkashingikai/kokugo/kokugo_kadai/iinkai_45/pdf/93390601_09.pdf ↩ ↩2 ↩3

  13. NHK 放送文化研究所 編. 『NHK日本語発音アクセント新辞典』. NHK 出版, 2016. ISBN 9784140113455. ↩ ↩2 ↩3 ↩4

  14. 国立国会図書館. レファレンス協同データベース. 「文部省が明治33（1900）年から明治41（1908）年まで，俗に『棒引き仮名遣い』を採用したことが…」. https://crd.ndl.go.jp/reference/detail?page=ref_view&id=1000186458 ↩ ↩2 ↩3

  15. 小学館. 「棒引き仮名遣い」. コトバンク (『日本国語大辞典』『精選版日本国語大辞典』 entry). https://kotobank.jp/word/%E6%A3%92%E5%BC%95%E3%81%8D%E4%BB%AE%E5%90%8D%E9%81%A3%E3%81%84-628439 ↩ ↩2 ↩3