白字

báizì

NounChinese character incorrectly written or read aloud in place of another character

Examples

这个字我写成了白字，应该是这样写的。Zhège zì wǒ xiě chéng le báizì, yīnggāi shì zhèyàng xiě de.I misspelled this character; it should be written like this.

About MyCidian

MyCidian is a free, web-based Chinese-English dictionary designed for native English speakers.

MyCidian is heavily inspired by Jisho.org, my favorite Japanese dictionary website. One of my biggest priorities when designing MyCidian was convenience. I wanted looking up a word to be quick, simple, and actually useful for someone who is learning Chinese.

This project is still ongoing and will only get better. It's free and staying that way. Web only, with no app planned. Ad-free for now; if that ever changes, signing in will remove them.

Some decisions

Recent lookups and your Study List stay in this browser.: MyCidian never receives which words you save, only an anonymous count of star clicks. Your list won't follow you to another device. Clearing site data deletes it, and some browsers, like Safari, may delete it if you don't visit for a while. Export your Study List to keep a copy.
Every alternate form, not a guess: One simplified character can stand for several traditional ones — 发 is both 發 (to send) and 髮 (hair). Picking one to display would look cleaner and lose the distinction.

Where the data comes from

CC-CEDICT: The core dictionary.
Make Me a Hanzi: Stroke order, radicals and character composition.
HSK 3.0 word lists: Proficiency-level browsing, by HSK 3.0 level. A word the 3.0 lists leave out keeps its HSK 2.0 level.
Example sentences: Over 70,000 checked sentences. About 51,000 were written by Qwen3.8-27B running locally; the rest (about 20,000) are older Claude Haiku examples. They were checked by Claude Sonnet, with a native-speaker spot check (94% natural in a 50-sentence sample). A few hundred older, unchecked Haiku examples are still shown where no checked sentence exists.
Claude Haiku: An enrichment pass grouping senses, plus a separate one tagging archaic and literary usage.

Built by Zeb Barrett

Credits

MyCidian is built on several openly licensed datasets. Every one currently in use is credited below, by name, with the licence it ships under.

CC-CEDICT: The core dictionary — headwords, pinyin and English definitions. Source, licensed CC BY-SA 4.0.
Make Me a Hanzi — dictionary data: Character definitions, derived from Unihan and CJKlib. Source, licensed LGPL 3.0.
Make Me a Hanzi — stroke-order graphics: Stroke order animation, radicals and character decomposition, derived from the Arphic PL KaitiM GB and PL UKai fonts. Source, licensed under the Arphic Public License.
hanziDB.csv: Character frequency ranking. Source, licensed MIT.
HSK 3.0 word lists: Proficiency-level browsing, compiled from the public HSK 3.0 standard by two open-source projects: complete-hsk-vocabulary and hsk30, both licensed MIT.
wordfreq: Word frequency ranking, used to put the commoner word first when several entries share a meaning. Compiled by Robyn Speer at Luminoso (version 3.1.1). Source, licensed CC BY-SA 4.0. MyCidian has changed the data: the frequencies are converted to a rank and matched to CC-CEDICT headwords.
The sources behind wordfreq's Chinese list: Wikipedia; OPUS OpenSubtitles 2018, from the OpenSubtitles project; SUBTLEX-CH, freely available data created by Marc Brysbaert et al. (Cai & Brysbaert, 2010); NewsCrawl 2014 and GlobalVoices news text; Google Books Ngrams 2012; web text from OSCAR; Twitter; and the word list distributed with the Jieba word segmenter.