Kana Chart & Quiz
An interactive hiragana and katakana chart with all three romanization systems side by side, and a drill that accepts whichever one you were taught.
Kana Chart and Quiz
How to Use the Kana Chart
Pick a script
Start with hiragana. It carries the grammar, so it opens up reading much sooner than katakana does.
Choose your romanization
Hepburn if you are learning from an English-language textbook; Kunrei-shiki if you are working from Japanese-published material. The gold-bordered cells are the ones where the systems disagree.
Tap any kana
The strip below the chart shows all three romanizations at once, plus a note when there is something genuinely awkward about that character.
Switch to the drill
Add the dakuten and yōon groups as the basic set gets easy, then the extended katakana once you are reading loanwords. Whichever romanization you were taught is accepted — the drill will tell you which system your answer came from.
Why Two Alphabets, and Why Nobody Agrees How to Spell Them
Japanese writes the same set of sounds twice. Hiragana and katakana cover an identical inventory — 46 basic characters each, plus the modifications — and a Japanese sentence uses both at once, alongside kanji. That looks like redundancy until you see what the choice of script is doing: hiragana handles the grammar and native vocabulary, katakana handles loanwords, onomatopoeia and emphasis. The script itself carries information that English has to signal with italics or capitals.
Both came from the same source. Before either existed, Japanese was written with Chinese characters used purely for their sound values, a system called man’yōgana. Hiragana are those characters written fast and cursive until the shape simplified. Katakana are fragments — a scribe taking one component out of a character as shorthand, in the margins of Buddhist texts. One script is a whole word worn smooth; the other is a piece broken off.
The romanization problem, and why it is not pedantry
A learner meets this within a week of starting, usually as confusion rather than as a stated fact. One book spells the kana し as "shi". Another spells it "si". A third writes つ as "tsu" where a Japanese-published workbook writes "tu". Nothing is wrong in any of them: they are three different standards, all current.
Hepburn spells for the English-speaking ear, which is why it is on passports and road signs and in nearly every textbook published outside Japan. Kunrei-shiki spells for the structure of the language — it keeps the T row regular as ta/ti/tu/te/to, because in Japanese those genuinely are one series, however unlike English spelling that looks. Kunrei is ISO 3602 and the system Japanese primary schools teach. Nihon-shiki goes further still and transcribes the letters rather than the sounds, which makes it the only one you can mechanically reverse.
A learner who writes "si" for し has not made a mistake. They have been taught the system their Japanese teacher was taught. A quiz that marks them wrong is testing which textbook they own.
That is the reason this chart shows all three and the drill accepts any of them. The differences are not scattered noise, either — they cluster on nine kana, and knowing which nine is genuinely useful, because those are exactly the characters where two sources will appear to contradict each other.
What the chart quietly admits
Two characters on this chart, ゐ and ゑ, were removed from ordinary writing by a cabinet directive in 1946. Another, を, survives for a single grammatical job and is pronounced "o" despite being romanized "wo" in Nihon-shiki and on a great many charts online. The empty cells in the Y and W rows are not layout accidents; they are where sounds fell out of the language and left the grid with holes in it. A chart that silently deletes all of this is easier to read and slightly false.
The part of the chart that is only thirty-odd years old
Everything above describes a syllabary settled centuries ago. But katakana has a job the classical table cannot do: it writes foreign words, and foreign words contain sounds Japanese does not have. There is no cell for fa, or vi, or che. So combinations were improvised — a full-size kana followed by a small vowel, ファ from フ plus a small ア — and in 1991 a cabinet notification, 外来語の表記, wrote the common ones down.
This matters more than its footnote-sized reputation suggests. If katakana’s main job is loanwords, then these are not an advanced topic to reach later: they are in the first hundred katakana words anyone reads. ファイル, パーティー, チェック, ディズニー, カフェ. A learner who has drilled the 46-cell katakana table and nothing else will look at フォーク and read four characters where there are three sounds.
They are kept as a separate group here, and switched off by default, for the same reason ゐ and ゑ are: mixing them into the gojūon would misrepresent what the gojūon is. They are also katakana only — there is no hiragana ファ, because hiragana is not the script that writes loanwords — and they carry a single romaji column, because Kunrei-shiki and Nihon-shiki genuinely have nothing to say about them. Two blank columns would look like missing data; one column with the reason attached is the truth.
The ヴ row is worth a second look, because it is the one part of the notification that is quietly losing. ヴ writes a v sound, and Japanese has no v — most speakers read ヴァイオリン and バイオリン identically. The B row usually wins in practice, and in 2019 Japan’s foreign ministry went as far as respelling two countries’ official names to drop ヴ altogether. The row is on the chart because you will read it; it is not on the chart because you should prefer it.
Learning Japanese from Southeast Asia
This is not a niche pursuit in this region. The Japan Foundation’s periodic survey of Japanese-language education has for years placed Indonesia among the largest populations of Japanese learners anywhere in the world, with Thailand, Vietnam and the Philippines also carrying substantial numbers — driven by secondary-school language programmes, manufacturing and tourism ties, and labour mobility schemes that make Japanese a direct employment credential rather than a hobby.
That changes what a kana chart is for. A learner in Jakarta or Hanoi is frequently working towards a JLPT sitting with a fixed date attached, often self-studying alongside work, and often from material published in Japan — which means Kunrei-shiki spellings turn up in their textbook and Hepburn spellings turn up in every app they use to drill. Being told one of them is wrong is worse than unhelpful; it makes a learner distrust the source that was right. Showing both, and naming which is which, costs nothing and removes a real obstacle.
11 Facts About Kana and Romanization
Both syllabaries descend from man'yōgana — Chinese characters borrowed purely for their sound. Hiragana are cursive whole characters; katakana are fragments, often a single component lifted out of one.
The 1946 cabinet directive on modern kana usage retired ゐ and ゑ from ordinary writing. They are still legible to any adult reader, which is why they appear on this chart rather than being quietly deleted from history.
Hepburn romanization is named after James Curtis Hepburn, an American physician and missionary whose Japanese–English dictionary popularised it — the third edition, in 1886, is the one that fixed the spellings.
Kunrei-shiki is not a rival hobby project: it is ISO 3602, and it is the system Japanese primary schools teach. A learner writing "si" for し has been taught correctly.
Gojūon means "fifty sounds", but the modern chart has 46 basic kana. The grid was designed around a fuller set, and the empty cells in the Y and W rows are where sounds dropped out of the language.
を survives in modern Japanese for exactly one job — marking the object of a verb — and it is pronounced "o". A chart that prints a bare "wo" is teaching a pronunciation nobody uses.
ん was a late addition. Classical Japanese had no syllable-final nasal, and the character had to be invented once Chinese loanwords brought the sound in.
じ/ぢ and ず/づ once sounded different. Their collapse into two sounds is the yotsugana merger, and it is why only Nihon-shiki still bothers to spell ぢ as "di".
The small っ writes no sound of its own — it holds the following consonant for one extra beat. Japanese speakers count it as a full mora, which is why きって has three beats, not two.
Katakana carries loanwords, onomatopoeia and emphasis, so a Japanese page uses it roughly the way English uses italics — the shape of the script itself is doing work the words are not.
The combinations that write foreign sounds — ファ, ヴィ, チェ, ティ — were standardised by a 1991 cabinet notification, 外来語の表記. They are the newest part of the writing system on this page, and the part a katakana learner meets first.
Frequently Asked Questions
- Because there are three, they are all in current use, and they disagree. Hepburn is what appears on passports, road signs and most textbooks written for English speakers. Kunrei-shiki is ISO 3602 and the system taught in Japanese primary schools. Nihon-shiki is the strict one that transcribes letters rather than sounds. Charts that print a single romaji column are making a choice and presenting it as a fact, which is confusing precisely when a learner meets a second source that spells it differently.
- Yes. "si" is the Kunrei-shiki and Nihon-shiki spelling, and it is what a child in a Japanese school is taught to write. "shi" is Hepburn. Both are correct romanizations of the same kana, so the drill accepts either and tells you which system you just used. Being marked wrong for using the system your teacher taught you is the most common complaint about kana quizzes, and it is entirely avoidable.
- If you are learning Japanese as an English speaker, Hepburn — it is what your textbook, your dictionary and every romanized sign in Japan will use, and it maps better onto English spelling instincts. Kunrei-shiki is worth recognising rather than adopting, because you will meet it in Japanese-published material and in older academic writing. The deeper answer is that romanization is scaffolding: the sooner you read かな directly, the sooner the question stops mattering.
- They were removed from ordinary writing by the 1946 reform on modern kana usage. You will still meet them in pre-war text, in some proper nouns and occasionally as a deliberate stylistic choice, so they are shown for completeness — but they are excluded from the drill by default, because a beginner should not spend repetitions on two characters they will never need to write.
- Hiragana, almost always. It carries the grammar — the particles and verb endings that hold a sentence together — so it unlocks reading much earlier. Katakana is mostly loanwords, which feels easier because you can often guess the meaning, and is in practice harder to retain because several characters look alike (ソ and ン, シ and ツ). Learning hiragana first also means the katakana chart is already familiar in shape.
- A dakuten is the pair of small marks that voices a consonant: か ka becomes が ga, さ sa becomes ざ za. The handakuten circle turns the H row into P — は ha becomes ぱ pa. They are not new characters to memorise so much as a modification applied to ones you already know, which is why they are a separate group here rather than mixed into the main table.
- By the standard rules, including the awkward ones. The small っ doubles the consonant that follows, except before ち, where Hepburn writes "tch" — まっちゃ is matcha, never maccha. ん takes an apostrophe before a vowel or y, so しんや is shin'ya and stays distinct from しにゃ shinya. Long vowels are marked with a macron rather than doubled: とうきょう is Tōkyō, not "toukyou", which is a keyboard input convention rather than a romanization.
- They are in the Extended group, which you can switch on with the other checkboxes. Those combinations were standardised by a cabinet notification in 1991 on writing loanwords, and since writing loanwords is what katakana mostly does, they are among the first katakana you will actually meet — ファイル file, ヴィオラ viola, チェック check, パーティー party. They are kept as their own group rather than mixed into the main table for two reasons: they are katakana only, because there is no hiragana ファ, and they carry one romanization column instead of three.
- Because Kunrei-shiki and Nihon-shiki do not define these combinations. Both systems are romanizations of the classical syllabary, written well before ファ and ヴィ were standardised, and neither has a spelling for them. Filling those two columns would mean inventing spellings and presenting them as standards, which is exactly the failure this tool exists to avoid — so the group says Hepburn only, and says why. One consequence worth noticing: Hepburn writes ティ as "ti", which is also the Kunrei-shiki spelling of ち. They never appear in the same table, so they never collide, but it is a good illustration of why one romaji column labelled "romaji" is not enough.
- Your display preferences are — which script, which groups, which romanization system — and they are kept in your own browser. Your score is not: it resets when you reload, on purpose. There is no account, nothing is sent anywhere, and no answer you type leaves the page.
- Because picking at random genuinely does show one character four times while another never appears, and learners reasonably read that as the tool being broken. Each card is drawn from the ones not yet seen in this pass, so a 46-card deck gives you all 46 before anything comes round again.
- Once the page has loaded, yes. The whole chart is a few kilobytes of data in the page itself — there is no dictionary to fetch, no API call behind the drill, and nothing that stops working on a slow connection.
Related News
You may be interested in these recent stories from our newsroom.
No related news yet for this tool. Our editorial team publishes new pieces every week.
Browse all news →Pick up where you left off
Stored only in this browser — never sent to our servers.