Chinese words used in English, ranked

English says tofu and kung fu so often that nobody hears them as foreign, reaches for feng shui and yin and yang when there is no English word for the idea, and says typhoon without the word very likely being Chinese at all. "Chinese" is doing a lot of quiet work in that last sentence, too: of the 21 loanwords below, only 7 actually come from Mandarin. Most came from Cantonese or Hokkien (also called Min Nan), two Chinese languages Mandarin speakers cannot understand by ear, by way of the ports and Chinatowns where English speakers actually met Chinese food and Chinese people. We ranked each word by how often it is said in 735 million words of English film and television subtitles, and traced which Chinese language it came from.

Updated 9 October 2026 · 8 min read

The words, ranked by how often English actually says them

The Chinese column gives the word in the language it actually came from — not always Mandarin. The Mandarin-rank column is the word's place among 85 million words of Mandarin subtitles, and it is empty for some rows on purpose: a Cantonese or Hokkien word can be written with ordinary Chinese characters without being a word Mandarin speakers ever say, which is exactly the fact the whole rest of this site's Shanghainese course turns on. Even so, 18 of these 21 words also turn out to be real, current Mandarin vocabulary. The 7 rows marked with an asterisk are two English words, not one — kung fu, feng shui, dim sum — and every half of each is also an ordinary English word or name on its own (chop, dim, sum, bok, chow), so a word count cannot isolate the phrase. The rank given there is a floor: the count of whichever compact spelling (kung-fu, dim-sum) is actually attested in writing, not of the far larger number of times the two plain English words happen to sit next to each other.

Only 3 of the 21 are common enough in English to rank in its top 10,000 words at all: tea, yang, and — of all things — ketchup.

Which Chinese language each one actually came from

• Mandarin (7 words): tai chi, feng shui, yin, yang and gung-ho all came from the national standard, along with kung fu (功夫, gōngfu), whose Mandarin meaning is just "skill, hard work" — no martial art required. Gung-ho is the strangest of these: it is short for a 1930s acronym, 工合 gōnghé ("Industrial Cooperative Society"), that a US Marine officer reinterpreted as "work together" and gave to his battalion as a motto. It is now too obscure even for Mandarin subtitles to say.
• Cantonese (7 words): the restaurant words — wok, dim sum, chop suey, wonton, bok choy — came overwhelmingly from Cantonese, the language of the Chinatowns of 19th- and 20th-century emigration, not from Mandarin, the language of Beijing. Kowtow is written the same way in both and could have come from either.
• Hokkien, also called Min Nan (4 words): ginseng, oolong and, as its own section below explains, tea itself.
• One more, chow mein, came from Taishanese specifically — a language in the same Yue family as Cantonese, and close to it, but not the same language.

Which Chinese language has the most speakers, and exactly how different Mandarin and Cantonese really are, is in Mandarin vs Cantonese; every branch of the family, by number of speakers, is sized up in the Chinese languages, by number of speakers.

Tea and cha: the same character, two different loanwords

Tea is Hokkien, not Mandarin — specifically the Amoy pronunciation, tê, that Dutch traders picked up at the port of Amoy and passed into English as thee, then tea. Say the Mandarin or Cantonese reading of the very same character, 茶, out loud instead and you get chá — which is where English's other, slangier word for tea, cha or char (as in "a cuppa char"), comes from, along with Hindi, Russian and Swahili chai. One Chinese character gave English two unrelated-sounding loanwords, because two different Chinese languages say it two different ways, and English happened to meet both.

Two words that probably are not Chinese at all

• Ketchup is "probably ultimately" Hokkien 膎汁 kê-chiap, "pickled fish sauce", by way of Malay — Wiktionary's own hedge — and modern Mandarin has no connection to the word whatsoever: the sauce is 番茄酱 fānqiéjiàng, built from entirely different characters.
• Typhoon is the word everyone assumes is Chinese and probably is not. Wiktionary traces it mainly to Ancient Greek tuphṏn by way of Arabic ṭūfān, carried into English by Portuguese sailors; a Sinitic origin, by way of a Cantonese phrase for "big wind", is noted only as something "sometimes thought", not established.

Meanings that moved once the word crossed into Mandarin too

A few of these loanwords also exist as ordinary Mandarin words — just not with the English sense.

• Bok choy is Cantonese for "white vegetable", 白菜. In Mandarin, 白菜 báicài usually means the larger Napa cabbage instead; bok choy itself is 小白菜 or 油菜.
• Chop suey is Cantonese for "mixed scraps", 雜碎. The same two characters in Mandarin, zásuì, mean giblets or entrails — not a stir-fried dish.
• Dim sum is Cantonese for "touch the heart", 點心. Mandarin uses the same characters, diǎnxīn, for any snack or dessert at all, not specifically this cuisine.
• Oolong is "black dragon", 烏龍. In current Mandarin slang, 乌龙 also means a blunder or an own goal in football — nothing to do with tea.
• Wok is the Cantonese word for the pan, 鑐. Mandarin's everyday word for the same pan is a different character, 锅 (guō); wok survives in Mandarin only as a rare, literary reading of a character nobody cooks with.

One word with a Shanghainese footnote

Mahjong is Mandarin, 麻將 májiàng — but Wiktionary's etymology does not stop there. It traces the word to 麻雀 máquè, "sparrow" (the tiles are said to clatter like sparrows chattering), with its final syllable built from a diminutive ending Wiktionary credits to Wu Chinese — the language family Shanghainese belongs to, and this site's one fully hand-written, fully voiced Chinese course besides Mandarin. Lychee has a cleaner split: it is a genuine doublet, with the spelling "lychee" reflecting the Cantonese reading of 荔枝 (lai6 zi1) and "litchi" reflecting the Hokkien one (lē-chi) — the same two characters, two Chinese languages, two English spellings.

Why this matters for a learner

None of this vocabulary is a head start on learning Mandarin: half of it is Cantonese or Hokkien, and even the Mandarin-origin words — gōnghé, the obscure acronym behind gung-ho — are not necessarily words a Mandarin course would ever teach. The useful list is the frequency-ordered one: the 1,000 most common Mandarin words, free, with pinyin on every one, and the Mandarin course that teaches them in that order — hand-written lessons with the grammar explained in English, reading, listening and writing sections, free on the web and working in any browser, including on a phone. The two-minute Mandarin vocabulary test estimates how many of them you already know, and is Mandarin hard to learn and how many words you need to speak Mandarin go further into what the language actually demands — tones and characters, mostly, not these loanwords.

The same exercise for a language English borrowed from very differently — through trade empires rather than emigration — is in Portuguese words used in English, and Italian words used in English measures another set of borrowings the same way, by how often each language actually says its own word.

Sources

• Lison, P. & Tiedemann, J. (2016). OpenSubtitles2016: Extracting Large Parallel Corpora from Movie and TV Subtitles. Proceedings of LREC 2016. aclanthology.org
• Dave, H. FrequencyWords: frequency lists from OpenSubtitles 2018 (English and Mandarin), content licensed CC BY-SA 4.0. github.com.
• Wiktionary, English etymology sections, checked 2026-10-09: kung fu, tai chi, feng shui, gung ho, yin, yang, tofu, mahjong, ginseng, lychee, wok, dim sum, chow mein, chop suey, wonton, bok choy, kowtow, oolong, tea, ketchup, typhoon.
• The word list, every origin and every meaning note are written by hand in scripts/frequency_extras.py (chinese_loanwords); the ranks and per-million figures are computed by scripts/frequency_curves.py and printed from its output rather than typed.