Chinese polyphone (多音字) lookup. Enter one character for its readings in pinyin and zhuyin, with meanings and example words for common ones; or paste text to see each polyphone with its in-context reading. No AI, browser-only.
Chinese Polyphone Lookup
How to use
Pick a mode
"Lookup character" shows the readings of one character; "Scan text" finds the polyphones inside a passage you paste.
Look up one character
In lookup mode, enter a single character (e.g. 行, 重, 的). The tool lists its readings in pinyin and zhuyin. 51 common polyphones also get a short meaning and example words; the rest show readings only.
Scan a passage
Switch to scan mode and paste a sentence or article. The reading appears above each polyphone. A character read two ways gets two rows, each with its snippet of your text and the other possible readings.
Learn from the examples
Every reading comes with example words — e.g. 行 as xíng (行走 walk) vs háng (银行 bank). Reading the examples is the fastest way to separate the readings.
Polyphones (多音字): When One Character Has Many Readings
A polyphone — 多音字 (duōyīnzì), also called 破音字 — is a single Chinese character that has two or more readings, where each reading usually carries a different meaning or part of speech. Polyphones are one of the most error-prone parts of learning Chinese, for native speakers and foreign learners alike. In the 2013 dictionary keyed to the mainland's official character list, 430 of the 3,500 level-one characters have more than one reading, about one in eight (counted from Unicode's Unihan transcription). Other lists keep different characters, so they give other rates. A handful, such as 行, 重, 长, 乐, 还, 差, 着 and 假, are the likeliest to be misread. For example, 行 read as xíng means "walk / okay / conduct" (行走 to walk, 不行 not allowed, 行动 action), but read as háng it means "row / trade / institution" (银行 bank, 行业 industry, 内行 expert). Likewise 重 as zhòng means "heavy / important" (重量 weight, 重要 important), while 重 as chóng means "again / repeat" (重复 repeat, 重新 anew). One character, different meanings.
Why polyphones exist
Multiple readings arise from three main sources. First, historical sound change and literary-vs-colloquial readings (文白异读): the same character is read differently in formal written speech versus everyday speech — e.g. 血 is xuè in formal reading, xiě in colloquial. Second, semantic split: a character took on several meanings, and the language assigned distinct readings to keep them apart — e.g. 乐 is lè for "happy" but yuè for "music." Third, a change of part of speech or grammar: the same character reads differently as a noun versus a verb — e.g. 长 is cháng ("length") or zhǎng ("to grow / leader"). Understanding these patterns helps you reason about an unfamiliar context far better than rote memorisation.
How to master polyphones efficiently
The most effective technique is "let the word fix the sound" (以词定音): rather than memorising a bare character, bind each reading to a few signature words. When you see 重, first ask: is it 重要 / 重量 (zhòng) or 重复 / 重新 (chóng)? The example words flash to mind and the reading settles itself. This tool's "Lookup character" mode is built for that: each reading of a common polyphone comes with example words, so you can form reading-to-word associations. The "Scan text" mode does the reverse and puts polyphones back into real sentences, so you can test your judgement against context. The tool runs entirely in your browser with no AI and no network calls.
"Let the word fix the sound; let the context fix the sense" — the working rule for polyphones.
The word decides the reading, not the character
How many polyphones there are depends on the list. Of the 8,105 characters in the 2013 dictionary, 597 have more than one reading. Taiwan's 1999 table (《國語一字多音審訂表》) lists 4,253 polyphones and keeps 957; its 2012 draft takes 1,263 down to 580.
着 is among the most complex common polyphones, with four readings — zhe (particle), zháo (catch fire), zhuó (wear), zhāo (a move) — each with its own job.
差 has four readings — chā (difference), chà (nearly), chāi (business trip), cī (uneven, in 参差) — spanning noun, adjective, verb, and reduplicative uses.
Many polyphones come from literary-vs-colloquial readings (文白异读): 血 is formal xuè but colloquial xiě; 壳 is formal qiào but colloquial ké.
Readings often mark part of speech. 长 as cháng is mostly adjective/noun (length); as zhǎng it is mostly verb/noun (grow, chief).
Names and places are polyphone hotspots. 重庆 is Chóngqìng, not Zhòngqìng, and scan mode reads it that way. Proper-noun readings often must be memorised separately.
还 is hái for "still / also" and huán for "return / restore". Paste 我还没还书 into scan mode to see the two readings separated.
The key to a polyphone is the word, not the lone character. Scan mode reads whole words first, and converts Traditional text to Simplified to look them up. A word it cannot find still falls back to the single character's default.
Readings can differ across the strait. 危 is wēi in the mainland and wéi in Taiwan's 1999 table; 企 is qǐ in the mainland and qì in Taiwan's 2012 draft. Mind the regional standard when checking a dictionary.
Frequently Asked Questions
-
A polyphone is a single Chinese character with two or more readings, where each reading usually maps to a different meaning or part of speech. For example 乐 is lè for "happy" but yuè for "music." In the 2013 dictionary keyed to the mainland's official character list, 430 of the 3,500 level-one characters have more than one reading, about one in eight.
-
"Lookup character" takes one character and shows its readings; 51 common polyphones also have meanings and example words. "Scan text" takes a passage and puts the reading above each polyphone, then lists one row per reading: a character read two ways gets two rows, and repeats of the same reading are counted on one row.
-
Scan mode reads by word through pinyin-pro's Simplified word list, adds a table of our own, and handles the particles 得 and 地 separately. Traditional text is converted to Simplified as a lookup key, so it also gets mainland readings, not Taiwan's. We test the page's 311 example words in both scripts: 303 match their printed reading and the other 8 need a sentence to decide, such as 穿着 and 朝阳. That is a self-check on bare words, not an accuracy rate. Rare collocations, names and ambiguous sentences can still come out wrong, so check a dictionary for anything that matters.
-
Three sources, and the page says which it used. RECATOOLS wrote the meanings and examples for 51 common polyphones. The other 546 are the characters the 2013 dictionary gives more than one reading, 597 in all, of which 430 are level one; those readings come from Unicode Unihan 17.0.0 (Unicode License V3), with no meanings. Scan mode's in-context reading comes from the open-source pinyin-pro library.
-
You can type either. A Traditional character is looked up as itself, then through its Simplified form, and the page says when it did. All readings follow mainland sources, though. Taiwan's Ministry of Education has its own standard: the 1999 version of 《國語一字多音審訂表》 is the one in force, and the 2012 revision is a draft. This tool carries neither, so check them directly for Taiwan readings.
-
No. A polyphone is one character with multiple readings; a variant character (异体字) is multiple glyph forms sharing one meaning and reading (e.g. 群/羣). This tool handles polyphones only, not variant-form merging.
-
No. All lookups and scans run locally in your browser with no server calls, and no AI or network service is used.
-
Scan mode flags only the 597 listed characters. pinyin-pro's own list is wider, but it includes many rare readings (it gives 不 and 信 a second reading each), so it is not used for flagging. Enter the character in "Lookup character" and the page shows pinyin-pro's reading and any others it lists, marked as unchecked.
-
The tool is meant for quick reading checks while learning. For formal publishing, exam-setting or proper-noun annotation, defer to an authoritative dictionary. To annotate a whole passage with pinyin, use the Hanzi to Pinyin converter.
Method & sources
How it computes
Lookup: the entry for the character typed, else the readings kTGHZ2013 gives it in Unicode Unihan 17.0.0, else a statement that it is not listed. Scan: pinyin-pro reads each run, Traditional via an OpenCC lookup key, then a table of words and two rules for the particles 得 and 地 correct it. One row per character and reading. Readings are mainland-sourced; Taiwan's tables are named on the page and not bundled.
What this tool implements
- Written 2026-11-01: 51 hand-written entries carry meanings and example words; 597 characters from Unihan, 430 of them level one, carry readings only, and the page says which list answered.
- Corrected 2026-11-01: the dataset was described as a hundred-plus characters when it held 50; the unsourced "one in six" is replaced by 430 of 3,500; 危 is wéi in Taiwan's 1999 table, not "also" wéi; 乐亭 Làotíng was cut because no primary source was opened.
- Tested: the kernel suite scans all 622 example words (311 in each script). 606 match their printed reading and 16 are declared ambiguous without a sentence. This is a self-check on bare words, not an accuracy rate.
Sources
- Unicode Consortium. UAX #38, Unicode Han Database (Unihan), Unicode 17.0.0: kTGHZ2013, readings as given in 通用规范汉字字典 (2013); kTGH, 通用规范汉字表. Unicode License V3. https://www.unicode.org/reports/tr38/
- Ministry of Education (Taiwan). 《國語一字多音審訂表》, status page: 88 年 3 月公告版「為現行規範,教科書審定等概以此版本內容為據」; the 12 Dec 2012 revision is a draft. https://language.moe.gov.tw/files/people_files/國語一字多音審訂表.pdf
- Ministry of Education (Taiwan). 《國語一字多音審訂表》, 1999, p. VIII: 「原列多音字(一字二音至一字八音)4253字」, 「審訂後仍為多音字(一字二音至一字六音)957字」; p. 363 lists 危 under ㄨㄟˊ.
- Ministry of Education (Taiwan). 《國語一字多音審訂表》 draft, 2012: 「總審訂字數為5,266字,其中多音字有1,263字」, 審訂後為多音字 580; 企 as ㄑㄧˋ.
- Ministry of Education (PRC). 《普通话异读词审音表(修订稿)》 notice, 2016-06-06, whose attachment reprints the 1985 table in its left column; 血 xuè/xiě and 壳 qiào/ké agree with that column. https://web.archive.org/web/20261020203234/http://www.moe.gov.cn/…
What can make this go out of date
- None at runtime: self-hosted pinyin-pro 3.28.1 (MIT) and OpenCC-JS 1.3.1 (MIT), and Unihan readings bundled in polyphone-data.js. Nothing typed leaves the browser.
Pick up where you left off
Stored only in this browser — never sent to our servers.