How to Learn Katakana (and Why It's Harder Than Hiragana)

A lot of learners give hiragana a solid week and katakana a weekend. Months later they read hiragana without thinking, and still stop dead at シ and ツ in the middle of a café menu.

Katakana has the same 46 basic sounds as hiragana, so on paper it should take the same effort. In practice it takes more. You see far less of it, because an ordinary page of Japanese is mostly hiragana and kanji with a katakana word here and there. The shapes are all straight lines and sharp corners, and several of them differ only by the angle of one stroke. And the words written in katakana are mostly borrowed from English, which sounds like it should help and often doesn't: コーヒー (koohii) doesn't look much like "coffee" until you've read it a few times.

The basic katakana chart

Each row is a consonant and each column a vowel, in the same order as the hiragana chart. If you already know hiragana, you know every sound here. Only the shapes are new.

Four characters don't match the pattern of their row: シ is shi, チ is chi, ツ is tsu, and フ is fu, a soft sound made with both lips that sits somewhere between "fu" and "hu". ヲ is pronounced o and is rare in katakana. ン is the only character that isn't a consonant plus a vowel.

Two small marks at the top right (dakuten) voice the consonant, so k becomes g, s becomes z, t becomes d and h becomes b. A small circle (handakuten) turns the h row into p.

ヂ and ヅ are almost never used in katakana words. The chart also extends with small ャ, ュ and ョ after an i-column character, which blend into one syllable: キャ (kya), シュ (shu), チョ (cho). ニュース (nyuusu) is news and シャツ (shatsu) is a shirt.

What katakana is used for

Katakana started as a shorthand. Monks in the Heian period took pieces of kanji to mark pronunciation in Buddhist texts, which is why the characters look like fragments: カ is the left side of 加, and イ is the left side of 伊. Today it has several jobs.

Loanwords are the big one. Most come from English, but not all of them. パン (pan, bread) came from Portuguese, アルバイト (arubaito, part-time job) is German Arbeit, and イクラ (ikura, salmon roe) is Russian ikra.

Emphasis works a bit like italics or capitals in English. A sign might write ダメ where a textbook would write だめ, just to make it louder, and manga often puts a robot's or a foreigner's stilted speech in katakana. Species names are written in katakana in scientific and educational writing, so a biology textbook has ヒト (hito, human) and ネコ where a novel would have 人 and 猫. Sound effects go either way. Many are katakana, and plenty are written in hiragana.

The look-alikes

This is where the time goes. Four groups cause most of the misreadings, and each has a tell.

シ and ツ

Both have two short strokes and one long one. In シ (shi) the short strokes are stacked one above the other on the left, and the long stroke is drawn from the bottom left up to the top right. In ツ (tsu) the short strokes sit side by side along the top, and the long stroke comes down from the top right to the bottom left. Hiragana helps here. し runs down the left side and hooks upward, and シ lines up along the left the same way. つ runs across the top and curves down, and ツ lines up along the top.

ソ and ン

The same idea with one short stroke. ソ (so) has its short stroke at the top, and the long stroke falls from the top right. ン (n) has its short stroke on the left, lying flatter, and the long stroke sweeps up from the bottom left. If the strokes look like they're hanging from a ceiling, you're looking at ソ or ツ. If they look like they're stacked against a wall on the left, it's ン or シ.

In print you can't see which way a stroke was drawn, so look at the short strokes instead. They're close to horizontal in シ and ン, and close to vertical in ツ and ソ.

ク, ケ and タ

ク (ku) is two strokes: a short slanted one, then a corner that sweeps down to the left. タ (ta) is ク with one extra short stroke across the inside. ケ (ke) has no corner. It has a flat horizontal bar, and its last stroke drops from the middle of that bar.

ウ, ワ and フ

フ (fu) is a single stroke, across and then down to the left. ワ (wa) adds a short vertical stroke on the left side. ウ (u) is ワ with a small tick standing on top, and hiragana う has a tick on top too.

Other pairs that deserve a second look: マ (ma) and ム (mu), コ (ko) and ユ (yu), ス (su) and ヌ (nu), ル (ru) and レ (re), チ (chi) and テ (te). Writing each pair by hand a few times does more than staring at them, because stroke order and direction are exactly what separates them.

The long vowel mark ー

Hiragana usually lengthens a vowel by adding another vowel character, as in おかあさん (okaasan, mother). Katakana uses a line. ー holds the vowel before it for one more beat. コーヒー is ko-o-hi-i, ケーキ (keeki) is cake, and スーパー (suupaa) is a supermarket.

Length changes meaning, so don't skip it. ビル (biru) is a building. ビール (biiru) is a beer. In horizontal text the mark looks a lot like the kanji 一 (one), but it almost always sits inside a katakana word, so context sorts it out. In vertical text it turns and runs vertically.

The other timing mark is small ッ. It doubles the consonant that follows, with a tiny pause before it: ベッド (beddo, bed), カップ (kappu, cup), チケット (chiketto, ticket).

Extra combinations for foreign sounds

Japanese has no native ti, fa or wi, so katakana grew extra combinations to get closer to foreign words. Each one pairs a full-size character with a small vowel.

ヴ is the odd one. It's ウ with dakuten, invented for the v sound, and it's used alone for vu or with a small vowel for va, vi, ve and vo. Most speakers pronounce it as b anyway, and the b spelling is very common, so you'll see both ヴァイオリン and バイオリン for violin.

Older loanwords were borrowed before these combinations caught on, and they keep the nearest plain sound. Team is チーム (chiimu), not ティーム. Radio is ラジオ (rajio). Vitamin is ビタミン (bitamin). That's also why ticket, in the section above, starts with チ.

Practice by reading real loanwords

Chart drills get you to the point where you can name each character. Reading speed comes from words, and loanwords make good practice because you can check your own answer. Sound the word out and see if something familiar falls out. Cover the right-hand column and try these.

A few habits of the system make decoding easier once you know them:

• Japanese syllables end in a vowel or ン, so extra vowels get added to break up consonant clusters. It's usually u, and o after t and d. That's how "street" becomes ストリート (sutoriito).
• L and r both land in the ラ row. ライト (raito) can be light or right.
• "Th" becomes s or z. サンキュー (sankyuu) is "thank you".
• Long words get clipped. パソコン is short for パーソナルコンピューター, and スマホ for スマートフォン.

Some katakana words look English and mean something else in Japan. マンション (manshon) is an apartment or condominium building, and コンセント (konsento) is an electrical outlet. You can't decode those. You learn them as vocabulary.

For daily practice, read every katakana word you pass: café menus, snack packaging, film titles, the credits of a game. Write your own name and your hometown in katakana. Ten minutes of that a day for a couple of weeks does more for reading speed than another run through the chart.

If you haven't done hiragana yet, start there. Our hiragana guide has the chart and a study order. For the look-alike pairs, flashcards with spaced repetition work well, since the ones you keep mixing up come back more often.

Common questions