Chinese word-formation (組詞) tool. Type one character and see the 2-, 3- and 4-character words that contain it, each group ordered by how often the word occurs, with CC-CEDICT pinyin.
Chinese Word Formation (组词)
Words and pinyin: CC-CEDICT (MDBG, release of 2026-05-24). Order: wordfreq by Robyn Speer (v3.2), which includes SUBTLEX-CH (Cai & Brysbaert, 2010). Both are licensed CC BY-SA 4.0, and this tool's word list is shared under the same licence.
How to use
Enter one character
Type a single Chinese character, such as 学, 水 or 火. If you type more than one, the first is used.
Read the three groups
The tool lists up to 30 two-character, 10 three-character and 8 four-character words that contain your character. Each group is ordered by how often the word occurs in the wordfreq corpus, most frequent first. The number beside each group reads "shown of the dictionary's total".
Read the pinyin
The pinyin is CC-CEDICT's reading of the whole word, so an erhua word is one syllable (哪儿 nǎr) and a neutral tone has no mark (学生 xuésheng). 一 and 不 are written in their dictionary tones, not as spoken before another syllable: 一会儿 is written yīhuìr and said yíhuìr.
Try another character
Tap a character below to switch. The page downloads only the one of its 16 word-list files that holds your character; what you type is not sent anywhere.
Word Formation (組詞): From Single Characters to Real Vocabulary
"Word formation" (組詞, zǔcí) is one of the commonest drills in Chinese literacy teaching: given a character, name words that contain it. For 学 (study) that might be 上学 (go to school), 中学 (secondary school), 学习 (to learn), 学生 (student), 学校 (school), 留学 (study abroad), and all six appear in this tool's two-character list for 学. Being able to form words from a character shows that a learner knows how it behaves in real words, not only what it looks like.
Why two-character words come first
Most Chinese words are two characters long. The word list of the new HSK exam syllabus (CTI, published 2025) has 11,000 entries, 10,896 distinct words once repeats are removed, and 78.5% of them are two characters long against 13.8% that are one character. That is a share of the entries in a word list, not of the words in running text. So the tool shows two-character words first and gives them the most room.
Where the words come from, and how they are ordered
Words and pinyin come from CC-CEDICT, the open Chinese-English dictionary (MDBG release of 2026-05-24, 124,945 entries). The tool takes its words of two to four characters and lists each one under every character it contains. Each group is ordered by wordfreq, Robyn Speer's word-frequency table, which combines film and TV subtitles (including SUBTLEX-CH), Wikipedia, news, books and web text. That is frequency in general text, not in a textbook, so under 上 the names 上海 (Shanghai) and 上帝 (God) rank high and 上课 (attend class) misses the first 30; you will find it under 课. Words wordfreq has no count for come last, in dictionary order.
The default results are filtered, and every filter reads a label CC-CEDICT itself gives. Words it tags vulgar, offensive or taboo, or glosses as a derogatory term for people, are left out. Senses tagged slang are dropped, and a word with no other sense is left out. Entries that only point to another spelling ("variant of …") are left out. Names of people, places and brands appear only when they are common (a wordfreq frequency of at least once per 100,000 words), and historical figures and dynasties dated with years never appear. Weekdays, months, festivals, temperature scales and continents are capitalised in CC-CEDICT too, but they are ordinary words and are listed as such: 星期天 (Sunday) is one.
Pinyin and Traditional characters
The pinyin is CC-CEDICT's reading written as one word: erhua joins the syllable before it (一点儿 yīdiǎnr), and a syllable starting with a, o or e takes an apostrophe (女儿 nǚ'ér). Where the dictionary gives two readings, both are shown, so 大夫 reads dàfū / dàifu. The word list is indexed by Simplified characters; if you enter a Traditional one (e.g. 學), OpenCC converts it to 学 for the lookup.
Who it is for, and its limits
It suits primary-school pupils and their parents, Chinese teachers and anyone learning Chinese. Try to think of words first, then use the tool to fill the gaps. The list is not complete: each group has a limit, words of five or more characters are not included, and a rare character may have no entry. Lookups run in your browser and the page keeps no record of what you type.
What 学, 水, 子 and 儿 do inside words
CC-CEDICT holds 589 words of two to four characters containing 学: 145 of two characters, 222 of three and 222 of four. The tool shows the 48 most frequent, with 大学生 (university student) and 留学生 (overseas student) among the three-character words and 北京大学 (Peking University) among the four.
One character can be read differently in different words. 乐 is lè in 快乐 (happy) and yuè in 音乐 (music); 行 is xíng in 旅行 (travel) and háng in 银行 (bank). The tool gives the reading of the whole word.
The tone shows whether 子 is a suffix. In 桌子 zhuōzi (table) and 孩子 háizi (child) it is toneless and adds nothing but the noun ending; in 君子 jūnzǐ (gentleman) it keeps its tone and its meaning, "person". Of the 121 words ending in 子 in the new HSK syllabus, CC-CEDICT reads 105 with a toneless zi.
儿 works the same way. In 哪儿 nǎr (where) and 一会儿 yīhuìr (a moment) it merges into the syllable before it as erhua; in 女儿 nǚ'ér (daughter) it is a syllable of its own and means "child".
Order matters: CC-CEDICT has 上学 (go to school) but no 学上.
Three- and four-character words are often more specific: under 水 (water) you will find 自来水 (tap water) and the idiom 水落石出 (when the water falls the rocks appear, the truth comes out).
Look-alike characters form quite different words, which is a good way to tell them apart: 己 / 已 / 巳 give 自己 (oneself) / 已经 (already) / 巳时 (the snake hour, 9 to 11 a.m.).
Frequently Asked Questions
-
Word formation means building words from a single character. Forming words for 学 (study) gives 上学, 学习, 学校, 留学 and so on. It is a basic literacy drill that turns single characters into usable vocabulary.
-
By length first (two, three, then four characters), then within each group by how often the word occurs in the wordfreq frequency table, most frequent first. wordfreq combines subtitles, Wikipedia, news, books and web text, so the order reflects general text, not a textbook. Each group shows at most 30, 10 and 8 words, and the number beside it says how many the dictionary has.
-
Yes. The word list is indexed by Simplified characters; if you enter a Traditional one (e.g. 學), the tool converts it to Simplified (学) with OpenCC before looking it up, and says so above the results.
-
It is CC-CEDICT's reading for each word, stored when the word list is built rather than guessed in the page. So a character with several readings is read by word: 行 is xíng in 旅行 (travel) and háng in 银行 (bank), and an erhua word is one syllable, 哪儿 nǎr. 一 and 不 are written in their dictionary tones (一会儿 yīhuìr); in speech they change.
-
Yes, using CC-CEDICT's own labels: words it tags vulgar, offensive or taboo, or glosses as a derogatory term for people, are left out, and senses tagged slang are dropped. The labels are the dictionary editors', not a word-by-word judgement of ours, so a word the dictionary has not labelled can still get through.
-
The list holds only CC-CEDICT words of two to four characters. If the dictionary has no such word for your character, or every one was removed by the filters above, the tool says so. Try a more common character.
-
No. The lookup runs in your browser. The page downloads one of 16 word-list files from our server, so the server sees which file was fetched, not the character you typed. The tool does not save your lookups in the browser.
-
No. Each group has a limit, words of five or more characters are not included, and the order follows general text, so a classroom word can miss the cut: 上课 (attend class) is not among the first 30 two-character words for 上, but it is listed under 课. If a word you expect is missing, look up its other character.
Method & sources
What this tool implements
- The order is a rule, not a judgement: frequency in wordfreq's Chinese table, which mixes subtitles, Wikipedia, news, books and web text. It reflects general text, not a textbook, and the page says so (上课 falls outside the first 30 words for 上 and is listed under 课).
- Measured once against the HSK 3.0 exam syllabus (CTI, 2025), which is not shipped: of its 163 multi-character level-1 words, the old lists showed 52 under every character they contain and the new ones show 147.
- The learner filter reads only CC-CEDICT's own labels: vulgar, (offensive), (taboo ...) or "derogatory term for" on any sense removes a word; slang senses and variant-only entries are dropped; names are kept only at Zipf 4 or more, and people or dynasties dated with years never; weekdays, months, festivals and continents, which CC-CEDICT capitalises, count as ordinary words.
- Pinyin follows the HSK syllabus's spelling of erhua (哪儿 nǎr) and writes 一 and 不 in their dictionary tones, as CC-CEDICT records them.
- Corrected 2026-11-03: "the most common words" (the lists were in code-point order), "hundreds of thousands of entries" (124,945), "about 70% of everyday text is two-character words" (no source; replaced by the HSK syllabus count, 78.5% of its 10,896 distinct words), Singapore and Malaysia described as using Traditional characters, "vocabulary size is the single strongest predictor of reading fluency" (no source), a pinyin-pro dictionary size the library does not state, and example words the tool did not show.
Sources
- MDBG. CC-CEDICT, release of 2026-05-24, 124,945 entries. Words and pinyin. CC BY-SA 4.0: https://www.mdbg.net/chinese/dictionary?page=cc-cedict
- Speer, R. wordfreq v3.2, Chinese table large_zh (commit 42233e6c). Word order. Data CC BY-SA 4.0: https://github.com/rspeer/wordfreq
- Cai, Q. & Brysbaert, M. (2010). SUBTLEX-CH: Chinese word and character frequencies based on film subtitles. PLoS ONE 5(6): e10729, https://doi.org/10.1371/journal.pone.0010729 (part of wordfreq's Chinese table)
- 中外语言交流合作中心. 中文水平考试 HSK 考试大纲, 2025-11 发布, 2026-07 实施. Counts only, for the two-character share and the coverage check: https://hsk.cn-bj.ufileos.com/3.0/新版HSK考试大纲1219.pdf
- OpenCC (opencc-js), MIT licence, for Traditional input and display: https://github.com/nk2028/opencc-js
What can make this go out of date
- None at runtime beyond the page's own files: the page fetches one of 16 data files and nothing the reader types is sent. The data is CC BY-SA 4.0 because of its sources, and the credit under the tool must stay.
Pick up where you left off
Stored only in this browser — never sent to our servers.