Of the 3,500 characters in the first level of mainland China's 2013 standard list, 89% share their exact Mandarin pronunciation, tone included, with at least one other character on that same level, by our count from Unicode's Unihan 17.0 data. Counting each character once, by the reading Unihan marks as most customary, a first-level character has 4.6 such twins on average, and against all 8,105 characters on the list 95.6% of the 3,500 have one. Read in Cantonese, by Unihan's customary readings for the same characters, the 3,500 spread over 1,340 toned syllables against Mandarin's 1,106, and 85% still have a twin.
China's standard dictionary sorts its characters into about 1,300 toned syllables
The list is the 《通用规范汉字表》 (《通用規範漢字表》), the Table of General Standard Chinese Characters, issued with State Council notice 国发〔2013〕23号. The notice says that once it is published, the earlier related lists 「停止使用」, cease to be used. Unicode's description of Unihan, UAX #38 for version 17.0, gives the list's three levels as 3,500, 3,000 and 1,605 characters, "those in the first level being more frequently used".
The readings come from the companion 2013 dictionary, 《通用规范汉字字典》 (《通用規範漢字字典》), as Unihan transcribes it in the field kTGHZ2013. Counting all 8,773 readings, the 8,105 characters use 409 syllables without tones and 1,296 with them, the neutral tone included. A toneless syllable is the pinyin alone (shi); a toned one adds the tone (shì). Our 1,296 matches the figure Unicode states for that field.
The 1983 second edition of the 《现代汉语词典》 (《現代漢語詞典》), in the field kXHC1983, gives 419 and 1,336 on our count, one fewer than Unicode's stated 1,337, a gap we could not explain. San Duanmu and Yan Dong's 2014 paper on homophone density puts Standard Chinese at 1,300 syllables with tones, citing Duanmu's 2007 work.
These are syllables some character uses in a given dictionary, a narrower thing than every sound Mandarin allows. The 1958 Hanyu Pinyin scheme marks four tones and leaves the neutral tone, 轻声 (輕聲), unmarked, with the example 妈 (媽) mā, 麻 má, 马 (馬) mǎ, 骂 (罵) mà and 吗 (嗎) ma. It gives no syllable count. On the 2013 list only 174 of the 409 syllables occur in all four full tones, and 35 occur in just one.
Level-one characters crowd unevenly onto ji, yi and shi
From here on, each character counts once, by the first value in Unihan's field kMandarin, the customary reading "preferred for zh-Hans (CN)". For 18 first-level characters, 了 and 的 among them, it gives the neutral-tone particle reading.
On that basis the 3,500 use 397 toneless syllables, with a median of seven characters each. Ji holds 52 first-level characters, yi 45 and shi 44. With tones, the 3,500 occupy 1,106 syllables, and half of those hold two characters or fewer; the busiest are shì and yì, tied at 22. At the other end, 375 first-level characters, or 10.7%, have a sound that no other first-level character shares. Measured against the whole 8,105-character list, only 155 have no twin.
The falling fourth tone carries 1,160 of the 3,500 characters. The dipping third tone carries 611, the fewest of the four full tones. The leader moves with the list: across all 8,105 characters, one reading each, yi leads with 131 and shi drops to about tenth.
Tone separates about two-thirds of same-syllable pairs
Among first-level characters, 25,087 pairs share a toneless syllable. The tone pulls 68.0% of those pairs apart, by our count on customary readings, and leaves 8,036 pairs of exact homophones. Across all 8,105 characters the tone separates 64.7% of same-syllable pairs.
These are counts of single characters; we did not measure words. A popular explanation holds that Chinese built two-syllable words such as 煤炭 (coal) to escape homophony among single syllables. Duanmu and Dong review that view critically and, testing it on the Chinese lexicon, report "no evidence for the correlation" between homophony and word length.
Cantonese splits Mandarin's shì nine ways
Of Mandarin's 8,036 exact-homophone pairs among first-level characters, 64.7% are pronounced differently in Cantonese. The Cantonese readings are Unihan's customary ones (field kCantonese) for the same 3,500 mainland-standard characters, in jyutping, a romanisation that numbers six tones. In reverse, 50.8% of Cantonese's 5,775 exact pairs come apart in Mandarin. Each reading crowds a different set of characters together.
The 22 first-level characters read shì fall into nine Cantonese syllables. 士 and 事 stay together as si6, 式 and 释 (釋) move to sik1, and 世 and 势 (勢) to sai3. 市 goes alone to si5, and 室 to sat1.
Cantonese spreads the 3,500 over 1,340 toned syllables, and a character shares its exact Cantonese sound with 3.3 others on average, against 4.6 in Mandarin. Two caveats apply. Jyutping counts endings in -p, -t and -k as part of the syllable, so in 223 of the 1,340 the ending already does some separating. And kCantonese gives one customary reading, so where a simplified form merges several traditional characters it follows one of them. Run on the 3,927 traditional forms of the first-level characters, with Unihan's Taiwan-preferred Mandarin, the direction holds.
An older count points the same way. A 2006 study by Ke, reproduced in Duanmu and Dong's paper, gives 1,125 syllables for Beijing and 1,367 for its Cantonese point, Guangzhou, over a list of some 3,000 common characters. Our 1,106 and 1,340 sit within 2% of those, on different data.
Extra readings divide some characters between several syllables
A character with more than one reading is a polyphone, 多音字. In the 2013 dictionary, 430 first-level characters are polyphones, and together they add 491 readings to the 3,500, or 14%. Editions also disagree: 妨 fāng and 哩 lǐ appear in the 1983 dictionary's transcription and not in the 2013 one. kMandarin also gives 25 first-level characters a second value, preferred for Taiwan. In about 11 of them the pronunciation itself differs, as with 识 (識), shì in Unihan's Taiwan value against shí, and 讽 (諷), fèng against fěng. That rests on Unihan alone, without Taiwan's Ministry of Education dictionary; Taiwan's school readings are covered in our zhuyin guide.
The Chinese homophones finder covers 19 crowded syllables and counts every 2013 reading of a first-level character, so it gives ji 56 characters where we give 52. Until 8 October 2026 the finder's table was typed by hand. Twelve listings carried readings the 2013 dictionary lacks, such as 果 under guó; a character search could match inside an example word, so 工 opened the shi table through 施工; and it held 282 of the 409 first-level readings in its syllables. The rebuilt table is generated from Unihan 17.0, lists all 409 readings, of 383 characters, and accepts either script and tone numbers such as shi4.
Number puns mostly pair near misses in Mandarin
The two best-known number puns are near misses. 四 sì (four) and 死 sǐ (death) differ only in tone, and 八 bā and 发 (發) fā only rhyme, in Cantonese as in Mandarin, as our Hong Kong car-plate guide sets out. Two other pairs behave differently. 六 liù and 禄 (祿) lù differ in Mandarin, but both are luk6 in Cantonese. 九 (nine) and 久 (long-lasting) match exactly in each: jiǔ in Mandarin and gau2 in Cantonese.
