Katakana

Katakana (カタカナ, written 片仮名, 'fragmentary kana') is one of the two Japanese syllabaries. Each basic letter stands for one mora, a beat of sound: a bare vowel (ア a, イ i, ウ u, エ e, オ o), a consonant plus a vowel (カ ka, キ ki, ク ku, ケ ke, コ ko), or the lone nasal ン n. The chart above is the gojūon ('fifty sounds') table that every learner memorises and every Japanese dictionary, phone book and karaoke index follows: five columns for a, i, u, e, o and rows for the consonants k, s, t, n, h, m, y, r and w. Four cells are read irregularly, シ shi, チ chi, ツ tsu and フ fu, and ヲ is pronounced o. The y and w rows have gaps, which is why the 'fifty sounds' come to 46 letters.

ヰ wi and ヱ we filled two of those gaps until the 1946 spelling reform retired them; the grid shows them in their old places because older charts, signs and texts still do. Both survive in brand names, ヱビス (Yebisu beer) and ニッカウヰスキー (Nikka Whisky), and in the names of old shops and shrines. ヲ itself is now almost only seen in all-katakana text such as pre-war documents, telegrams and the speech of robots in manga; in ordinary writing the object particle is the hiragana を.

Every katakana is a piece of a Chinese character. In the 9th century monks in the temples around Nara and Kyoto needed a fast way to mark Japanese readings between the lines of Chinese Buddhist texts, so they took a few strokes of a kanji used for its sound: ア from 阿, イ from 伊, ウ from 宇, カ from 加, ニ from 二, ハ from 八, ミ from 三, メ from 女, ツ from 川, ノ from 乃. That is why katakana are angular where hiragana (see the Hiragana collection), which grew from whole characters written in flowing cursive, are round, and why both sets have a letter for the same 46 sounds. For centuries each sound had several competing shapes; the one-letter-per-sound chart you see here was fixed by the Ministry of Education in 1900.

Dakuten, handakuten and small katakana: the full katakana chart

Two small marks extend the 46 letters. The dakuten ゛ (two short strokes at the top right) voices a consonant: カ ka becomes ガ ga, サ sa becomes ザ za, タ ta becomes ダ da, ハ ha becomes バ ba. The handakuten ゜ (a small circle) turns the h row into p: ハ becomes パ pa, ヒ becomes ピ pi. That gives 25 more letters, and ウ with a dakuten, ヴ vu, writes the foreign v sound: ヴァイオリン (violin, also spelled バイオリン), ヴィーナス (Venus). The single letters ヷ ヸ ヹ ヺ (va, vi, ve, vo) exist in Unicode but almost nobody uses them; the spellings ヴァ ヴィ ヴェ ヴォ with a small vowel are the ones you will see. ヂ and ヅ sound exactly like ジ and ズ and are rare in katakana words.

Small katakana change the letter next to them. Small ャ ュ ョ fuse with the kana before them into one syllable: キャ kya in キャンプ (camp), ジュ ju in ジュース (juice), チョ cho in チョコ (chocolate). Small ッ doubles the following consonant: カット (cut), ベッド (bed), サッカー (soccer); at the end of a word in manga, アッ, it shows a sound cut short. The small vowels ァ ィ ゥ ェ ォ build the sounds Japanese did not have, which is the main job of katakana: ファ fa in ファイル (file), ティ ti in パーティー (party), ウィ wi in ウィキペディア (Wikipedia), ディ di in ディズニー (Disney), フォ fo in フォーク (fork), ジェ je in ジェット (jet), チェ che in チェック (check), トゥ tu in タトゥー (tattoo), ツァ tsa in モーツァルト (Mozart). These spellings were standardised in the Cabinet's 1991 guidelines on writing loanwords.

ヶ and ヵ look like small ケ and カ but are really abbreviations of the counter 箇 (ka): 三ヶ月 san-kagetsu (three months), 一ヶ所 or 一ヵ所 ikkasho (one place), and in place names, where ヶ is read ga: 霞ヶ関 Kasumigaseki, 自由ヶ丘 Jiyūgaoka. Small ヮ belongs to old spellings such as クヮ. The newest small kana, 𛅕 (small ko) and 𛅤 𛅥 𛅦 𛅧 (small wi, we, wo, n), were added to Unicode in 2019 and 2022 for phonetic transcription and dialect texts; most fonts still show them as boxes, so keep them out of titles and messages.

The long-vowel bar ー (chōonpu) is the most used character on this page after the letters themselves. It stretches the vowel before it: コーヒー kōhī (coffee), ケーキ kēki (cake), ラーメン rāmen, スーパー sūpā (supermarket). In vertical text it turns upright. Always copy the real bar: a hyphen -, a dash — or the kanji 一 (one) look similar in some fonts but are different characters, so コ一ヒ一 will not be found by a search for コーヒー. The middle dot ・ separates the parts of a foreign name, ジョン・レノン (John Lennon), and the items of a list; ヽ and ヾ repeat the previous kana (ヾ with voicing) and belong to older texts; ゠ is the double hyphen that some reference works use for hyphens inside foreign names, where most people simply type =. ヿ is an old ligature of コト, 𛀀 a 9th-century katakana for a vowel that merged with エ, and 𛄠 𛄡 𛄢 the yi, ye and wu that 19th-century chart makers invented to fill the empty cells; the last four are for scholars and need a recent font.

What katakana is used for

Modern Japanese writes with three scripts in one sentence: kanji for the meaning of most words, hiragana for grammar and native words, katakana for everything that comes from outside or needs to stand out (the Japanese Characters collection shows all three together). Katakana write loanwords, gairaigo: コーヒー (coffee), テレビ (television), コンピューター (computer), アルバイト (part-time job, from German Arbeit), パン (bread, from Portuguese pão). They write foreign names and places, アメリカ, ロンドン, トム, and so your own name in Japanese will be katakana. They write sound effects and mimetic words, above all in manga: ドキドキ (heart pounding), ワンワン (a dog's bark), ニャー (meow), ガーン (the sound of a shock). Biologists write the names of species in them, ヒト (human), イヌ (dog), サクラ (cherry). And they work like italics or capitals for emphasis and slang, ダメ, ヤバイ, or for a robot, an alien or a foreigner speaking with an accent.

The division of labour is recent. Until 1946 laws, official documents and army orders were written in kanji mixed with katakana, not hiragana; the Meiji Constitution of 1889 and the Emperor's 1945 surrender broadcast both read that way. Primary schools taught katakana first: the 1933 first-grade reader opened with サイタ サイタ サクラ ガ サイタ ('they bloomed, they bloomed, the cherry blossoms bloomed'). Telegrams were sent in katakana alone, which is why the stock Japanese telegram is ハハキトク スグカエレ, 'mother critical, come home at once'. After the war hiragana took over the schoolbooks and the laws, and katakana settled into the roles above.

Online, katakana became faces. The shrug ¯\_(ツ)_/¯ smiles with ツ tsu, (ノ◕ヮ◕)ノ*:・゚✧ throws sparkles with ノ for arms and ヮ for a mouth, and ヽ(´ー｀)ノ raises ヽ and ノ as arms over a ー mouth. The Kaomoji Symbols collection has hundreds more built this way, and gamers put a lone ツ in usernames as a smiley.

Ainu katakana, Taiwanese kana and halfwidth katakana

Ainu, the language of the indigenous people of Hokkaido, is written in katakana, and its syllables can end in a consonant, which Japanese katakana cannot show. The 16 small letters ㇰ ㇱ ㇲ ㇳ ㇴ ㇵ ㇶ ㇷ ㇸ ㇹ ㇺ ㇻ ㇼ ㇽ ㇾ ㇿ solve that: a small letter at the end of a syllable is read as its consonant alone, and its vowel echoes the vowel before it. アイヌ イタㇰ aynu itak is 'the Ainu language', イランカラㇷ゚テ irankarapte is the greeting 'hello' (ㇷ゚, small fu with a handakuten, is a final p), ピㇼカ pirka means 'good, beautiful'. Ainu also writes a few sounds with a handakuten on an ordinary kana, such as ト゚ tu and セ゚ ce. These letters joined Unicode in version 3.2 (2002) as the Katakana Phonetic Extensions, and the Japanese system fonts on Windows, macOS, iPhone and Android include them.

Taiwanese kana are a closed chapter. While Taiwan was under Japanese rule (1895–1945) the colonial government published dictionaries, schoolbooks and newspapers in Taiwanese Hokkien written in katakana, with small marks beside the letters for the language's tones; the great Taiwanese–Japanese Dictionary of 1931–32 is the best known. Unicode 14.0 (2021) encoded those tone marks as 𚿰 𚿱 𚿲 𚿳 𚿵 𚿶 and the nasalized 𚿷 𚿸 𚿹 𚿺 𚿻 𚿽 𚿾 in the Kana Extended-B block. Hardly any font has them yet, so expect boxes outside specialist software. The language itself is in the Taiwanese Hokkien Characters collection.

Halfwidth katakana ｱ ｲ ｳ ｴ ｵ are the katakana of the first Japanese computers. The 1969 JIS X 0201 standard fitted 63 katakana and marks into single bytes beside the Latin alphabet, drawn in half a character cell so that they lined up with it; the dakuten and handakuten became separate characters, ｶﾞ ga and ﾊﾟ pa, two bytes each. They lived on in Shift_JIS, on pagers, early mobile phones and 8-bit and 16-bit games, and they still appear on receipts, ATM screens and the name fields of Japanese bank transfers, which often insist on 半角カナ. Unicode keeps them as compatibility characters (U+FF65 to U+FF9F) that normalise to the full-size letters. Do not use them in running text, where they have been bad manners since the days when they broke Japanese e-mail; use them when you want that receipt-and-pager look, ｶﾀｶﾅ, the same retro taste that the Fullwidth Text (Vaporwave Letters) collection serves from the other direction.

How to type katakana, and which letters are easy to mix up

With a Japanese keyboard set up on Windows (Microsoft IME or Google Japanese Input), type the word in romaji, ko-hi-, and press F7 to turn it into full-width katakana コーヒー, F8 for halfwidth ｺｰﾋｰ and F6 to go back to hiragana; the - key gives ー and the / key gives ・. On a Mac with the Japanese – Romaji input source, type the romaji and press Control+K for katakana (Control+J for hiragana, Control+L for full-width Latin, Control+; for halfwidth katakana). Small kana take an x or l in front: xa → ァ, xtu → ッ, xya → ャ, xke → ヶ, and vu gives ヴ. On iPhone and Android the Japanese keyboard offers the katakana spelling among its suggestions after you type the word. In Microsoft Word you can also type the code and press Alt+X: 30C4 Alt+X gives ツ. For a single letter, tapping it on this page is faster than any of these.

Four pairs of katakana trip up every learner because they differ only in the angle of the strokes. シ shi and ツ tsu: in shi the two short strokes lie flat and the long stroke sweeps up from the bottom left; in tsu they stand upright and the long stroke sweeps down from the top right. ソ so and ン n follow the same rule. Then ク ku, ケ ke and タ ta; ヌ nu and ス su; ワ wa, ウ u and フ fu; コ ko and ユ yu; チ chi and テ te.

Several katakana are near-twins of kanji, and a search or a spell-checker treats them as different characters: カ ka and 力 (power), エ e and 工 (craft), ロ ro and 口 (mouth), ニ ni and 二 (two), ハ ha and 八 (eight), タ ta and 夕 (evening), ト to and 卜, オ o and 才. Katakana ヘ he and hiragana へ are drawn almost identically but are two characters. And the full-size letters on this page are not the same as the Katakana block's halfwidth cousins: ア and ｱ are both katakana a, but a search for one will not find the other unless the software normalises them. Every full-size letter, mark and Ainu letter renders in the stock fonts of any phone or computer; only the small kana added since 2019, the archaic 𛀀 𛄠 𛄡 𛄢 and the Taiwanese tone marks need a recent Noto font, and even then not every app will show them.