Lesson 8 · カタカナ — the second script
Twenty-five new shapes, no new sounds
Today's chart is day one's chart with every character swapped out. The same five rows, the same five vowels down the columns, the same three irregulars — and not one sound you have not been saying since the first lesson.
Japanese runs on two syllabaries, not one. They spell exactly the same sounds; what differs is which words get written in which. Everything so far has been hiragana. This is the other one, and because the sound system is already learned, the entire job today is recognising twenty-five shapes.
1.The same chart, in the other hand
Here is the grid from lesson 1 with the katakana set beside each character. Read down the columns and the vowels are unchanged; read across the rows and the consonants are unchanged. Nothing about the sound system has moved.
| row | a | i | u | e | o |
|---|---|---|---|---|---|
| vowel | あ ア | い イ | う ウ | え エ | お オ |
| k | か カ | き キ | く ク | け ケ | こ コ |
| s | さ サ | し シ | す ス | せ セ | そ ソ |
| t | た タ | ち チ | つ ツ | て テ | と ト |
| n | な ナ | に ニ | ぬ ヌ | ね ネ | の ノ |
Trick 1 · Do not learn twenty-five characters. Learn twenty-five spellings of things you can already say
Every other lesson so far added sounds: new consonants, voiced rows, a contracted beat. This one adds none. When you meet カ the answer is not "work out what this says" but "which character I already know is this the other spelling of?" That is a much cheaper question, and it is the only question this chart asks.
Trick 2 · The three irregulars are the same three
If the grid were regular the s and t rows would give si, ti, tu. They don't — and they don't in exactly the same three places as before. シ is shi, チ is chi, ツ is tsu. There is no second list of exceptions to learn; it is the same list, worn by different characters.
2.Why the shapes look the way they look
Both syllabaries were made in the Heian period out of the same raw material: Chinese characters that had been pressed into service to spell Japanese sounds, called man'yōgana. What separates them is not which characters they started from but what was done to them.
| Made by | So they are | |
|---|---|---|
| ひらがな | writing the whole kanji in flowing cursive | round, joined, one continuous movement |
| カタカナ | taking one piece of the kanji and keeping it | straight, angular, separate strokes |
That one difference explains the whole page. な is the kanji 奈 written fast, so it curls and loops through four strokes. ナ is the top of 奈 taken cleanly and left straight, so it is two. Same ancestor, opposite treatment.
Trick 3 · Katakana are not fewer strokes — they are straighter strokes
It is tempting to assume the angular script is the quicker one. It is not. These twenty-five katakana take 62 strokes between them; the twenty-five hiragana take 57. Cursive writing joins strokes together, so つ and て are one stroke each where ツ and テ are three. What katakana buy you is that every stroke is a straight line or a simple hook — easier to get right, not fewer to make.
Fifteen are siblings. Eight are strangers. Two are arguments.
Because the two scripts were cut from the same stock of man'yōgana, some pairs really do descend from one character — and where they do, the family resemblance is inherited and can be trusted. But not all of them do. Fifteen of these twenty-five pairs share an ancestor; eight came from completely different kanji and have no reason to look alike at all; and for two of them the sources do not agree.
| Pair | Katakana from | Hiragana from |
|---|---|---|
| ア あ | 阿 | 安 |
| イ い | 伊 | 以 |
| エ え | 江 | 衣 |
| ケ け | 介 | 計 |
| サ さ | 散 | 左 |
| ス す | 須 | 寸 |
| タ た | 多 | 太 |
| チ ち | 千 | 知 |
The remaining two are worth knowing about, because they are where the scholarship itself stops agreeing. ツ has no settled ancestor at all: the Nihon Kokugo Daijiten lists 川, 州, 津 and 闘, and つ is unsettled in the same way. ニ is unsettled differently: sources split evenly between 仁 and 二, and since に definitely comes from 仁, whether this pair are siblings is exactly what is in dispute. The shape comes out the same either way — 仁 is 亻 plus 二 — so the writing is settled even though the descent is not.
Trick 4 · Check whether a pair is related before you go looking for a likeness
This is the difference between a structural hook and a made-up one. コ really is こ with its right-hand ends joined, because both are 己 — lean on that. サ and さ also look a bit alike, and it means nothing: 散 and 左 are unrelated characters and the resemblance is an accident. Building a memory on the second kind is how you end up confidently writing the wrong letter. Tap any character in the next section and its card says which kind it is.
3.Write them, hear them, use them
Tap any character below. It draws itself in the correct stroke order — numbered, one stroke at a time — says itself aloud, names the hiragana that shares its sound and the kanji it was cut from, and gives three real words that use it, each glossed in English and in Arabic.
Each box carries the hiragana above and the katakana beneath, with the romaji shown once. The single romaji is the point: on lesson 5's chart the two characters in a box had different sounds and an arrow between them. Here they have the same sound, so there is no arrow and nothing to convert.
ア イ ウ エ オ カ キ ク ケ コ サ シ ス セ ソ タ チ ツ テ ト ナ ニ ヌ ネ ノ
The interactive chart needs JavaScript. The 25 characters are listed above in row order, and the katakana stroke chart has all of them drawn out.
Practise this way, not that way
Watch the animation once, then write the character on paper from memory before replaying it. Copying while looking feels productive and teaches you almost nothing; the effort of recalling the shape is what makes it stick. With katakana there is a second trap: because the shapes are simple, a wrong one still looks plausible. Recall it, then check — do not trace.
4.The pairs that get misread
Simple shapes made of straight strokes are quick to write and easy to confuse. Which pairs actually cause trouble is not a matter of opinion: a 2019 writing study tested all 45 katakana on 430 Japanese children in school years 1–3 and counted the errors.4 Its finding is blunt — ツ and シ lead the list, and around ten in every 134 children were still getting them wrong in year three.
| Pair | What actually separates them |
|---|---|
| シ ツ | Direction, not shape. The most-missed pair in the language, by a distance. See below. |
| ソ ン | The same direction rule with one mark instead of two. ン is not on today's sheet; the rule you learn now will settle it when it arrives. |
| ス ヌ | Touch or cross. On ス the second stroke only meets the first; on ヌ it cuts straight through and out the other side. |
| ク ケ | The top right. ケ's horizontal juts out to the right at the top; ク turns down instead. |
Two more that get listed in beginners' material and are worth a smaller share of your attention: ク against タ (one extra slash, three strokes against two), and ク against ワ and コ against ユ, both of which wait for the second half of the chart. ク/タ is worth knowing but is not what trips people: neither character appears anywhere in that study's error data.
Trick 5 · For シ and ツ, read the stroke direction
Both are two short marks and one long sweep, and printed side by side they are nearly the same picture. What separates them is the direction the pen travels, which the finished shape only hints at:
- シ — shi: the two marks come in from the left, roughly horizontal, and the long stroke is pulled upward from the bottom left. Its hiragana し also ends by sweeping up.
- ツ — tsu: the two marks come down from the top, roughly vertical, and the long stroke is pulled downward to the left. Its hiragana つ also runs left along the top.
Read off a finished letter, there are two independent giveaways and they point the same way. Where the strokes start: all three of シ's begin in a column down the left-hand side, while all three of ツ's begin in a row along the top. How the two short marks are angled: shallow, close to horizontal, on シ; steeper, close to upright, on ツ. And the same test, applied to one mark instead of two, is what will later separate ソ from ン — シ and ン are one family, ツ and ソ the other.
5.Read them cold
Everything above this line you could do while looking at the chart. This is the part that finds out whether the shapes are actually in your head: a character appears, you name its sound, four options, by tapping or by pressing 1–4. There is deliberately no play button on the question — hearing it is the answer. Get it right and the drill moves straight on. Get it wrong and it stops, tells you the hiragana that shares the sound, how the strokes go, and every word from the cards above with that character inside it.
The three wrong answers are not random, and after section 4 you can guess where they come from. They are drawn first from the pairs this lesson has flagged — シ against ツ, ス against ヌ, ク against ケ — then from the same row, and only then from anywhere. A multiple-choice question whose wrong answers are obviously wrong measures nothing.
One round is all twenty-five, shuffled. Miss some and it offers you those again on their own. Nothing is saved anywhere — this is a mirror, not a record.
The drill needs JavaScript. All twenty-five characters, their sounds and their words are in section 3 above.
6.What katakana is actually for
Hiragana and katakana spell the same sounds, so which one a word is written in is never about pronunciation — it is a label on the word. Seeing katakana tells you something about a word before you have finished reading it.
- Loanwords — foreign words taken into Japanese (gairaigo), which is by far the commonest use. Also foreign place names and foreign personal names: the United States is アメリカ.
- Onomatopoeia — words that are sounds.
- Technical and scientific terms, including the scientific names of animal and plant species and of minerals. This is why ネコ, キツネ and スイカ appear on the cards above in katakana even though they are ordinary Japanese words, not borrowings — but see the caution below, because this one is routinely overstated.
- Company names, which is how a firm is told apart from the surname it shares.
- Signs and advertising.
- Emphasis — doing the job italics do in English.
Where point 3 gets overstated — and it usually does
"Animals and plants are written in katakana" is most of the way to true and is worth being careful with. Japan's Agency for Cultural Affairs sets the actual rule for official writing:5 as an ordinary word, an animal or plant name is written in kanji if the character is on the standard list and in kana if it is not — so 犬 stays 犬. Katakana is the convention for the scientific name. The Ichthyological Society of Japan states it outright for fish: species names below the rank of order are to be written in katakana.6
So イヌ is closer to "Canis familiaris" than to "dog". Reading it as a plain synonym is the mistake.
And the reason is not that biology likes katakana. Before the war, ordinary Japanese was written in kanji and katakana, and species names were set in hiragana to stand out from it. After the post-war script reforms flipped the everyday mixture to kanji and hiragana, species names flipped too — to katakana. It is a contrast device: the name is written in whichever script the surrounding text is not.
Trick 6 · Sound it out in English before you look it up
Because most katakana words are borrowed, reading one often means recognising a word you already have. テレビ is terebi is television. ネクタイ is nekutai is necktie. スーパー is sūpā is supermarket. Say the syllables out loud and let your ear finish the word — it succeeds far more often than it has any right to, and it is the single fastest reading skill in Japanese.
It is not a guarantee. ケース is case, but the sense narrows to a container. And point 3 above means plenty of katakana words are not borrowed at all.
One word this course was always going to reach
コーヒー — kōhī — reached Japanese from Dutch koffie, and Dutch got it, through a chain of European languages, from Arabic قَهْوَة (qahwa).2 The Japanese card above and the Arabic gloss beside it are, historically, the same word arriving from two directions. Most loanwords on this page record only their most recent donor — English or Dutch — but this one is traced the whole way.
The one thing today's sheet does not give you is the long-vowel mark ー, which katakana uses constantly to hold a vowel for a second beat: it is the difference between コヒ and コーヒー. It appears in the words on the cards above because they would be misspelled without it, but it is not part of the twenty-five, and neither are the small kana or the marks. Those arrive with the rest of the chart.
7.Sources
- Tofugu, “Learn Katakana” — the clearest free walkthrough of the set, and it is organised around exactly the confusion pairs in section 4. Read this one before the next class.
- Wiktionary, Japanese kana entries — the man'yōgana ancestry for all fifty characters in section 2, taken from each character's own Etymology section and cross-checked against the Japanese Wikipedia 字源 entries. Also the route コーヒー took from قَهْوَة. The two sources disagree in two places and Wiktionary is followed in both; the reasoning is recorded with the data.
- Wikipedia, “Katakana” — the six uses listed in section 5.
- 森田安徳 (2019), 「小学1〜3年児童のカタカナ書字の誤りに ついて」, 神戸親和女子大学 児童教育学研究 38 — a writing test of all 45 katakana on 430 native children in school years 1–3, with the errors counted. This is what section 4 is ranked by, and it is why シ/ツ gets a callout while ク/タ gets a sentence: the first two dominate its error table and the second two do not appear in it at all.
- 文化審議会建議 「公用文作成の考え方」 (令和4年), Agency for Cultural Affairs — the rule for animal and plant names in official writing: standard-list kanji as an ordinary word, katakana as a scientific name. This is the source that makes “animals are written in katakana” too strong a claim.
- 日本魚類学会, 「魚類の標準和名の 命名ガイドライン」 (2020) — a learned society writing the katakana convention down as an explicit rule, §3.1.1. Its own preamble notes the field had operated on unwritten custom until then.
- KanjiVG, © Ulrich Apel, CC BY-SA 3.0 — every stroke order and direction on this page. It is the data behind the handwriting tools in Jisho.org and Tagaini Jisho, and it is what settles シ against ツ: prose can describe the difference, but the path data records which end of the stroke the pen starts from.
- Jisho.org, serving JMdict/EDICT — every reading and gloss, and the commonness tag that decided which words earned a place. It refused ウニ, ケニア and ノルウェー, and it caught that ナス in katakana is network-attached storage rather than an aubergine.
- NHK World — Easy Japanese — free lessons with native recordings, for checking any word on this page against a person rather than a speech synthesiser.
- Madinah Arabic and Almaany — the Arabic glosses and their harakat.
- Katakana stroke chart — all twenty-five drawn out with every stroke numbered, each beside its hiragana and its ancestor, laid out to print. Its companion is the hiragana stroke chart.