How to Use a Chinese Dictionary: Four Ways In

Every dictionary you have ever used before assumed one thing: that you know how the word sounds. You heard it, or you read it in an alphabet that told you. Then you looked it up in the order the alphabet dictates.

A Chinese character tells you very little about its sound and quite a lot about its meaning. So the problem splits in two, and the two halves need different tools. If you heard it, you have a sound and no shape. If you read it, you have a shape and no sound. Almost every difficulty a learner has with dictionaries comes from using the tool for the wrong half.

There are four doors. Learn all of them; you will use all four in a normal week.

First, a distinction that matters: 字典 and 词典

A 字典 (zìdiǎn) is a dictionary of characters. A 词典 (cídiǎn) is a dictionary of words — and most Chinese words are two characters or more. 《新华字典》 is the standard small character dictionary; 《现代汉语词典》 is the standard reference for words.

This matters constantly. 电 means electricity and 脑 means brain; 电脑 (diànnǎo) is a computer. Looking up the two characters separately gives you the etymology, not the meaning. If you met the word in a text, you want a 词典.

Door one: pinyin, when you know the sound

The fastest route when it works. Every dictionary has an 音序 index — alphabetical by pinyin syllable, then by tone, then by stroke count within a tone.

The trap is the homophone. Type shi and you will meet 是, 十, 事, 时, 市, 试, 室, 使, 石, 识… dozens of characters across four tones. Two things narrow it fast:

• Know the tone. It cuts the field by roughly three quarters, which is one more reason to learn tones as part of the word rather than as an afterthought.
• Search the whole word, not the syllable. If you heard shíjiān, search shijian, not shi. Two-syllable searches are nearly unambiguous.

If you are unsure of the tone, most digital dictionaries let you search without one. If you are unsure of the initial — you heard something between x and sh, or between q and ch — search both. That confusion is diagnostic, and looking up both is itself a listening lesson.

Door two: the radical, when you have the shape

This is the classical method, it is still how paper dictionaries are organised, and it is worth learning even in an age of cameras — because it teaches you to see characters as assemblies rather than pictures.

The procedure has three steps:

1. Identify the radical (部首, bùshǒu), the component under which the character is filed.
2. Count the radical's strokes and find it in the 部首检字表, the radical index at the front.
3. Count the remaining strokes — the whole character minus the radical — and find your character in that sub-list.

A worked example. 湖 (lake). The radical is 氵, three strokes, the water radical. Remaining component 胡 is nine strokes. So: radical index, three-stroke section, 氵; then within 氵, the nine-stroke sub-list; there is 湖, with its reading hú and a page reference. Twelve strokes in total.

Another. 简. The radical is ⺮ (bamboo), six strokes. What is left is 间, seven strokes. Thirteen in all, read jiǎn.

Where it goes wrong

The radical is often not the component you expect. Three that catch everyone:

• 相 is filed under 目 (eye), not 木 (tree).
• 和 is filed under 口 (mouth) in the Kangxi system, not 禾 (grain).
• 期 is filed under 月, not 其.

When the radical is genuinely unguessable, dictionaries provide a 难检字表 — an index of hard-to-look-up characters, arranged purely by total stroke count. Use it without embarrassment; it exists because the editors knew.

Door three: total stroke count

The fallback when even the radical defeats you, and the reason stroke order is not a calligraphic nicety. You cannot count strokes reliably unless you know how the character is written.

The rules you need are few:

• top to bottom, 从上到下
• left to right, 从左到右
• horizontal before vertical, 先横后竖 — 十 is two strokes, the horizontal first
• left-falling before right-falling, 先撇后捺 — 人
• outside before inside, 先外后内 — 同
• close the box last, 先里头后封口 — 国: the enclosing frame is finished only after what is inside it
• 辶 is written last, after everything it wraps — 这, 边, 进

And a handful of forms that look like more strokes than they are: 阝 is two strokes wherever it appears, 辶 is three, 女 is three, 马 is three, 及 is three.

Door four: handwriting and the camera

On a phone, the practical answer to "what is this character" is usually to draw it. Handwriting input panels are forgiving — they tolerate imperfect shapes — but they are far more accurate when your stroke order is right, because the recogniser uses the sequence and direction of strokes, not only the finished picture. This is the hidden reason to learn stroke order even if you never write by hand.

Camera and OCR lookup handles printed text: point the phone at a menu, a sign or a page and the characters are segmented and read for you. Pop-up dictionary readers do the same inside a browser or a document — you hover or tap and the entry appears. These are genuinely transformative for reading, and they carry one risk: it becomes possible to read for an hour without learning anything, because the answer arrives before you have tried to retrieve it. Give yourself a two-second rule. Try to recall first. Then tap.

Reading the entry once you get there

An entry is more than a gloss, and the parts learners skip are the parts that make the word usable.

Multiple readings. Many characters are 多音字 (duōyīnzì) with different pronunciations for different meanings:

• 行: xíng (to be OK, to walk) / háng (a row, a firm — 银行 yínháng, a bank)
• 长: cháng (long) / zhǎng (to grow, senior — 校长 xiàozhǎng)
• 觉: jué (to feel — 觉得) / jiào (sleep — 睡觉 shuìjiào)
• 乐: lè (happy — 快乐) / yuè (music — 音乐 yīnyuè)
• 重: zhòng (heavy) / chóng (again — 重新)
• 还: hái (still) / huán (to return something — 还书)

If the sentence does not make sense with the reading you assumed, check whether the character has another one. This single habit resolves a surprising share of reading dead-ends.

Part of speech and measure word. A good 词典 tells you that 消息 takes 条, that 建议 can be a verb or a noun, that 帮忙 is a verb-object compound and therefore splits: 帮我一个忙.

The example sentences. Read them. A gloss tells you what a word means; the examples tell you what it goes with, which is the part you cannot guess. 提高 goes with 水平 and 质量; 增加 goes with 数量 and 收入. No English translation of either word will tell you that.

A twenty-minute exercise

Take a page of Chinese you have not read. Pick ten characters you do not know. Look up five by radical in a paper or app-based radical index, timing yourself, and five by handwriting input. Write down for each: the radical, the total stroke count, the reading, and one word the character appears in.

Do that three times and radical lookup stops being an academic ritual. You will start seeing 氵 and knowing the character has something to do with water before you know anything else about it — which is exactly the kind of half-knowledge that makes reading possible.