How to read Chinese characters

Chinese characters look like thousands of unrelated drawings. They are closer to a construction kit: a few hundred recurring parts, combined by rules, in which most characters carry one hint about meaning and one about sound. Once you see the parts, the job changes from memorising pictures to recognising familiar pieces.

The second thing that makes reading easier than it looks is that you already read this way. Fluent readers of English do not spell words out letter by letter. They recognise whole shapes.

You don't read strokes or letters. You recognise words.

Tihs sntence si wrng in nearly evry wrod, and you read it anyway. Reading Chinese works on the same principle, with denser shapes.

How Chinese characters are built

Around 100 CE the scholar Xu Shen described six categories of characters, the 六書 (六书) liùshū. They are still the standard way to explain how characters work.

Category	Chinese	What it is	Examples
Pictographs	象形 xiàngxíng	Stylised drawings of things	日 rì, sun; 月 yuè, moon; 山 shān, mountain; 木 mù, tree
Simple indicatives	指事 zhǐshì	Abstract ideas shown by a sign	一 yī, one; 上 shàng, above; 下 xià, below; 本 běn, root
Compound ideographs	會意 (会意) huìyì	Two meanings combined into a third	休 xiū, to rest (a person by a tree); 林 lín, woods (two trees)
Phono-semantic compounds	形聲 (形声) xíngshēng	One part for meaning, one part for sound	媽 (妈) mā, mother; 河 hé, river; 清 qīng, clear
Loan characters	假借 jiǎjiè	A character borrowed for a word that sounded alike	來 (来) lái, "to come", originally a kind of wheat
Derived cognates	轉注 (转注) zhuǎnzhù	Related characters with related meanings; the definition has been disputed for centuries	考 kǎo and 老 lǎo

The first three categories get all the attention in beginner books, because the pictures are charming. They are also the minority. Well over 80 per cent of characters are phono-semantic compounds.

Take 媽 (妈) mā, mother. The left side, 女 nǚ, woman, tells you the field of meaning. The right side, 馬 (马) mǎ, horse, is there only for its sound. In 河 hé, river, the three drops of 氵 say "water" and 可 kě supplies a rough rhyme. The sound hint is often loose, since pronunciation has shifted over two thousand years, but a loose hint is still better than none.

Characters are also not words. Most modern Chinese words have two characters, and one character serves in many words. 電 (电) diàn, electricity, gives you 電話 (电话) diànhuà, telephone; 電腦 (电脑) diànnǎo, computer; 電影 (电影) diànyǐng, film; and 電梯 (电梯) diàntī, lift. Every character you learn is repaid several times.

How many Chinese characters do you need to know?

Fewer than the size of the dictionary suggests. Large dictionaries list tens of thousands of characters, most of them rare variants and antiques. Frequency is extremely uneven, as the linguist Jun Da's character frequency list, counted on a large corpus of modern Chinese texts, shows:

Most frequent characters	Share of all characters in the texts
100	about 42%
500	about 76%
1,000	about 89%
1,500	about 95%
2,000	about 97%
2,500	about 98.5%
3,000	about 99%

Two official lists point the same way. The mainland's Table of General Standard Chinese Characters (2013) holds 8,105 characters, of which the 3,500 in its first tier are those meant for basic education. Taiwan's standard list of common characters has 4,808.

The figures need one honest caveat. Covering 89 per cent of the characters on a page does not mean understanding 89 per cent of the text, because meaning lives in words, and unknown words are often made of known characters. Kungfucius's working numbers: with about 1,000 characters you can read graded material and guess your way through messages and signs; 2,500 to 3,000 is where newspapers and novels become reading, not decoding.

You do not read strokes, you recognise words

Here is a little experiment. Look at this line:

It says "Can't read this". It is set in Electroharmonix, a typeface whose Latin letters are drawn to look like Japanese katakana. Westerners can read it fairly easily after a second of surprise. Japanese readers have real difficulty with it, because their eyes keep trying to read the shapes as kana, and as kana the line is nonsense. The glossary entry on Electroharmonix tells the story.

Now read this:

u cann eeezili reed dis sheeeeet!

Not a single word in that sentence exists, and yet you read it without trouble. The reason is the same in both cases: we do not read letters. We recognise words and meaning.

It is the same with Asian languages. Nobody reading Chinese counts the strokes of 謝 or checks each line of 語. Over time you simply recognise words and meaning without seeing the exact strokes, just as you did with the two lines above. That is why reading a lot matters more than writing each character a hundred times.

Components and radicals

The practical unit of learning is the component, not the stroke. 清 has eleven strokes, but only two components: water on the left and 青 qīng on the right. A learner who knows both has almost nothing new to memorise.

Components repeat in families. 青 qīng lends its sound to 清 qīng (clear), 請 (请) qǐng (to invite), 情 qíng (feeling) and 晴 qíng (sunny), while water, speech, heart and sun separate the meanings. Meet one member and the others are half-learned.

A radical is one particular component: the one a dictionary files the character under, usually the meaning part. A short list of common ones is worth learning early, and the radicals page has it. Do not memorise all 214 before you start reading. That is preparation as a form of delay.

Traditional and Simplified characters

There are two standard sets. Simplified characters are used in mainland China, Singapore and Malaysia; Traditional characters in Taiwan, Hong Kong and Macau. Many everyday characters are identical in both, and most of the differences follow regular patterns: 言 becomes 讠 on the left of a character, 門 becomes 门, 馬 becomes 马.

Kungfucius's advice, set out on the Simplified vs Traditional page: start with the set used where you will use the language, and learn to read both for at least the 1,000 most common characters. Reading both is far less work than it sounds. Writing both by hand is optional.

Read first, write later

This is where Kungfucius parts company with many classrooms. The order that works is: ears, then eyes, then, if you insist, the hand.

1. Listen first. Build a base of spoken words through easy dialogues, with Pinyin as support. Chinese children know thousands of spoken words before they meet their first character. A character attached to a word you already know sticks; a character attached to nothing slides off.
2. Read along with audio you already understand. Follow the text of dialogues you know by ear. You are no longer learning words, only matching shapes to them, which is a far smaller task.
3. Type. Pinyin input lets you write real Chinese as soon as you can recognise characters: you type the sound and choose the character from a list.
4. Start the comic-book phase. Children move from sounding out letters to recognising whole words by reading piles of easy, entertaining material. The Chinese equivalent is graded readers and simple stories with audio. Volume matters more than perfection: recognising 80 per cent on a read-through is a good start.
5. Use flashcards for the stragglers. Put the characters you keep failing to recognise into a spaced-repetition deck. Kungfucius's rule of thumb is that flashcards should take about a tenth of your study time, not more.

What about handwriting?

Kungfucius's view is admittedly controversial: learn the basics of handwriting in a few weeks, then mostly stop. Learn the stroke-order rules (top before bottom, left before right, outside before inside, close the box last). Then write ten characters a day, ten times each, for about twenty days. That is enough to understand how characters are put together and to copy any character legibly.

After that, type. Many adults in China and Taiwan write little by hand after leaving school and reach for the phone when a rare character escapes them. The exceptions are real: some courses and some exam formats require handwriting, so check before you skip it entirely.

Tools for learning characters

• Pleco: the dictionary app. Look characters up by Pinyin, by drawing them or with the camera, and see their components and both script forms.
• Skritter: a paid app that drills handwriting and stroke order on the screen. Useful for the handwriting weeks, or if writing is a goal in itself.
• Anki: free flashcards with spaced repetition, best used for characters met in your own reading.
• Graded readers: Du Chinese offers short levelled texts with audio; book series such as Mandarin Companion and Chinese Breeze tell longer stories with a few hundred characters.

One dictionary, one source of easy reading and one flashcard tool are enough. Collecting apps is not a reading method, however much it feels like one.

A realistic plan

These stages are Kungfucius's estimates for a learner with about an hour a day, not measured results. Adjust them without guilt.

Stage	Rough time	Characters you recognise	Main activity
Sound first	Weeks 1 to 4	A few dozen, by accident	Listening, Pinyin, tones; notice characters, do not study them
First characters	Months 2 to 4	About 300	Read along with known dialogues; learn 20 to 30 common components
Handwriting basics	20 days, somewhere in months 3 to 6	No change	Ten characters a day, ten times each; then start typing
Easy reading	Months 4 to 12	About 1,000	Graded readers with audio; flashcards for stragglers only
Both scripts	Year 2	About 2,000	Longer graded texts, subtitles; learn to read the other character set
Native material	Year 3 onwards	2,500 to 3,000	News, novels, messages from friends, with a dictionary for the gaps

Start this week: choose one beginner dialogue you already understand by ear, open its text, and read along three times. Then look up five of its characters in Pleco and see which components they share. That is the whole method in miniature.