How do Chinese characters work? Meaning, sound and radicals

Chinese characters work by writing units of meaning rather than single sounds: each character stands for a morpheme, the smallest piece of a word that carries meaning, and is read as one syllable. A few began as pictures, but most are built from two parts, one hinting at what the character means and one hinting at how it sounds. Learning to read means learning a few hundred of those parts and how they combine.

In short

• A character usually writes one syllable with one meaning, so the script records words and their sounds together.
• Most characters are compounds of a meaning part, often the radical, and a sound part; true pictures are a small minority.
• A literate reader knows about 3,000 to 4,000 characters, out of tens of thousands in the largest dictionaries.

Are Chinese characters pictures?

Only a few are, and even those no longer look much like what they show. The oldest Chinese writing that survives is on oracle bones, ox shoulder blades and turtle shells used for royal divination by the Shang dynasty from about 1250 BC, at what is now Anyang. Some of its signs were drawings: 人 (rén) "person", 山 (shān) "mountain", 日 (rì) "sun", 木 (mù) "tree". Over three thousand years of brushes, carving and print, those drawings became the strokes written today, and a modern reader recognizes 山 because they learned it, not because it looks like a range of peaks.

A second small group combines meanings directly. 休 (xiū) "rest" puts a person beside a tree. 林 (lín) "grove" is two trees, and 森 (sēn) "forest" is three. These are memorable, which is why textbooks lead with them, but they give a false picture of the system. In one analysis of the 3,500 characters most used in Standard Chinese, summarized in Wikipedia's survey of the script, pure meaning signs made up only about 5%. Chinese writing is one of the four writing systems scholars accept as invented independently, alongside Mesopotamian cuneiform, Egyptian hieroglyphs and Maya script, and it is the only one of them still in use.

How does one character carry both meaning and sound?

Most characters are phono-semantic compounds: one component points to the meaning, and another points to the pronunciation. The word for "mother" shows the idea. Spoken Mandarin already had the syllable ma in it, and the character 马 (mǎ) already existed, meaning "horse". Writing "mother" as a horse would be confusing, so scribes added 女 (nǚ) "woman" on the left. The result, 妈 (mā), reads as "the woman word that sounds like ma". Nobody is meant to see a horse in it.

The same trick runs right through the script. The three short strokes 氵 are a squeezed form of 水 "water", and they mark 河 (hé) "river", 湖 (hú) "lake" and 流 (liú) "flow". In 湖, the right side 胡 (hú) gives the sound exactly; in 河, the right side 可 (kě) gives it only roughly. A set of characters that share one sound part is called a phonetic series, and reading one series shows how much a learner gets for free:

Character	Pinyin	Meaning part	Meaning
青	qīng	(the sound part alone)	blue-green
清	qīng	氵 water	clear, of water
晴	qíng	日 sun	clear, of weather
情	qíng	忄 heart	feeling
请	qǐng	讠 speech	to ask, please
睛	jīng	目 eye	eyeball

Step by step: 青 supplies a syllable close to qing; the left side says which field of meaning the word belongs to. A learner who knows 青 and the five meaning parts can make a strong guess at all six characters, and in reverse, a reader meeting 晴 for the first time can guess both "something about the sun or weather" and "probably said qing". Tones and the first sound may differ, as 睛 (jing) shows, so the hint narrows the guess rather than settling it.

Analyses differ on how large this class is. In the count above it covered 58% of the 3,500 common characters, and counts across whole dictionaries put it above 80%, because rare characters were made almost entirely this way. Either way it is the largest class by far.

What is a radical?

A radical is the component a dictionary files a character under. Before alphabetical order was possible, a dictionary needed some way to find a character, so lexicographers grouped characters by a shared part. Xu Shen's Shuowen Jiezi, written around AD 100, used 540 such headings. Mei Yingzuo's dictionary of 1615 cut them to 214 and ordered the characters under each by their remaining strokes, and the Kangxi Dictionary of 1716 made that set of 214 the standard. Its roughly 47,000 entries are still why people speak of "the 214 radicals".

Because most characters are phono-semantic, the radical is usually the meaning part: 女 for 妈 and 姐 "elder sister", 氵 for the water words, 讠 for words about speech. But radicals are a filing system, not a theory of meaning, and some are only shapes chosen to give a character somewhere to live. Mainland Chinese dictionaries for simplified characters use a revised list of 201 headings, published in 2009.

Why does the sound hint so often miss?

The hints were accurate when the characters were made, and the spoken language has moved for two or three thousand years since. 江 (jiāng) "river" pairs 氵 with 工 (gōng), which sounds little alike in Mandarin today but was much closer in Old Chinese. The same character can also sound different in Cantonese, Shanghainese or Japanese, because the script was never tied to one pronunciation.

That independence is a strength and a cost. Speakers of Chinese varieties that can't understand each other's speech can read the same newspaper. But Mandarin has only about 400 distinct syllables without tones, so many characters share a sound: 妈 mā "mother", 麻 má "hemp", 马 mǎ "horse" and 骂 mà "scold" differ only in tone, and many characters share the very same syllable and tone. Writing them differently keeps meanings apart that speech leaves to context. Pinyin, the official spelling in Roman letters adopted in 1958, records the sound, tones included, but doesn't replace the characters.

How many characters are there, and how many do you need?

There is no single total, because it depends on what counts as a separate character. The Kangxi Dictionary has about 47,000 entries, many of them rare or variant forms of the same character. The Unicode standard, which gives every character a number for computers, encoded more than 100,000 Chinese characters by 2025, most of them rare names or historical forms. Older encodings built around ASCII's 128 symbols couldn't hold them at all.

Daily reading takes far fewer. China's 2013 List of Commonly Used Standard Chinese Characters holds 8,105, and its first tier of 3,500 covers ordinary use. Studies of literacy put an educated reader's active knowledge at 3,000 to 4,000 characters, and a specialist in literature or history at 5,000 to 6,000. The work is front-loaded: once a learner knows the common components, each new character is a combination of known parts, which is why methods like spaced repetition suit it well.

Characters are also written stroke by stroke in a fixed order, generally top to bottom and left to right. The character 永 (yǒng) "forever" is the traditional teaching example, said to contain all eight basic strokes.

What is the difference between simplified and traditional characters?

Simplified characters are standard forms with fewer strokes that the People's Republic of China adopted in the 1950s and 1960s: a draft scheme in 1956, and a list of 2,235 characters in 1964. Most were not invented; they were shorthand forms already used in handwriting, or ancient forms. So 馬 became 马, 媽 became 妈, 門 "door" became 门 and 書 "book" became 书. A second round in 1977 met resistance and was withdrawn in 1986.

Mainland China and Singapore use simplified characters. Taiwan, Hong Kong and Macau use traditional ones. The structure is the same in both: 媽 and 妈 are still woman plus horse, and a reader of one set learns the other far faster than a beginner learns either.

Characters also spread beyond Chinese. Japanese writes with kanji, 2,136 of them on the government's list for general use since 2010, mixed with two syllabaries. Korea once wrote with hanja and now uses its own alphabet, hangul, almost entirely. Vietnam wrote formally in Chinese characters for centuries before a Latin-based alphabet replaced them in the 20th century. How alphabets got their start, by a different route, is told in where the alphabet came from, and how scripts fit into speech generally in what language is.

Check yourself

• Without looking, explain how 妈 "mother" is put together, and why it contains a horse.
• Why does a sound component often give only a rough guess at the pronunciation today?
• What is a radical for, and why isn't it always the part that gives the meaning?

Sources

• Chinese characters — Wikipedia
• Radical (Chinese characters) — Wikipedia
• Kangxi Dictionary — Wikipedia
• Notice of the State Council on publishing the List of Commonly Used Standard Chinese Characters — State Council of the People's Republic of China, 2013
• Unihan database and charts — Unicode Consortium
• Chinese Writing — World History Encyclopedia