Simplified vs Traditional Chinese: the history, the differences, and which to learn

What's the difference between Simplified and Traditional Chinese?

They're two standard ways to write one language, with the same words, grammar and pronunciation. What changes is the shape of some characters. "To study", 学(xué), is 學 in Traditional, and "to speak", 说(shuō), is 說. "Person", 人(rén), is the same in both.

Simplified characters, 简体字(jiǎntǐzì), are the forms the People's Republic of China standardised from 1956 on. Traditional characters, 繁體字(fántǐzì), are the older forms, still the standard in Taiwan, Hong Kong and Macau.

Taiwan calls them "standard characters", 正體字(zhèngtǐzì). That's the label on the registration form for its national Chinese test, the TOCFL.

So "Mandarin or Simplified?" isn't a real choice. Mandarin is a spoken language, and you can write it in either script. Cantonese is a different spoken language, and in Hong Kong it's written in Traditional characters too. Our Mandarin vs Cantonese guide covers that choice.

One thing the labels hide: switching characters doesn't switch vocabulary. "Taxi" is 出租车(chūzūchē) in Beijing, but usually 計程車(jìchéngchē) in Taiwan.

A converter that only swaps characters turns 出租车 into 出租車: correct Traditional, but not the usual Taiwan word. OpenCC, the open-source converter most apps use, keeps a separate Taiwan phrase list for exactly this reason.

Over a third of common characters differ

More than a third of the 1,000 most frequent characters have a different Traditional form. The exact share depends on which standard you compare against:

Source: our count, with characters ranked by frequency in SUBTLEX-CH, Cai and Brysbaert 2010. How we measured.

Those 1,000 characters make up 94% of everyday text in the film-and-TV subtitle corpus we used to rank them. Here's how the share holds as you go further down the list:

Why two numbers? The Taiwan figure also counts two kinds of difference that China doesn't call simplification:

• Variant choices. In a separate 1955 decision, China kept one form among several variants. It kept "at", 于(yú), where Taiwan writes 於.
• Printed shapes. Some characters differ only slightly in print, like "not have", 没(méi), which Taiwan prints as 沒.

A learner switching scripts meets both kinds, so the Taiwan figure is the better guide to the work involved.

A typical page changes by about a quarter

On a real page, the share is lower. The most frequent characters are the ones least likely to change: the possessive 的(de), "is", 是(shì), "I", 我(wǒ), and "you", 你(nǐ), are the same in both scripts.

Characters that always change make up 26% to 27% of the subtitle corpus, depending on the standard.

When we converted all of SayMei's graded stories, 25.7% of the characters changed.

Just five characters made up 23% of all the changes:

The share barely moves as you climb the HSK levels:

Source: character lists from the new HSK syllabus; differences are our count.

Simplified does save ink: about 19% fewer strokes in everyday text, by our count on Unihan's stroke data. The savings pile up on the most frequent characters:

Simplification made handwriting official

Mostly, simplification made official the shortcuts people already wrote by hand. Then it applied each simplified part to every character that contains it.

The first scheme simplified hundreds of characters and dozens of components. In a joint notice of 7 March 1964, China ruled that most of them also apply when they appear inside other characters.

Here's the usual classification of the methods, as Wikipedia's article on simplified characters summarises it:

The last row is why switching is less work than the percentages suggest. Most of China's 1964 General List was built that way, from a short list of simplified parts:

Source: General List, 1964.

Once you know that the "speech" part, 讠, is 言(yán), and the "metal" part, 钅, is 金(jīn), dozens of characters convert themselves. Explore 80 common pairs, grouped by how they were simplified:

One Simplified character can be two Traditional ones

Simplification merged characters that sounded alike, so going back needs the whole word. 发 stands for both "to send out", 發(fā), and "hair", 髮(fà):

• "to discover", 发现(fāxiàn), becomes 發現
• "hair", 头发(tóufa), becomes 頭髮

China's 2013 conversion table splits 96 groups like this. Its official explainer warns that converting "dipper", 斗(dǒu), blindly gives 北鬥星 for the Big Dipper, instead of 北斗星(Běidǒuxīng).

Half of them, 48 groups, involve characters among the 1,000 most frequent, by our count, so learners meet them early:

Good converters read the whole word, and most of the time they get it right. We tested the one SayMei uses for its Traditional view, opencc-js, on all 166 of our graded stories.

It got about 1 character in 1,260 wrong, by our audit. The mistakes cluster in a few patterns:

• A longer wrong word matches first. In "will it come clean after washing?", 洗了还能干净吗, the converter reads 能干(nénggàn), "capable". So it writes 能幹淨 instead of 能乾淨.
• No helpful neighbour. In "the noodles are tasty", 面很好吃, nothing next to 面 says "noodles", so it stays 面, "face". In our noodle-shop story, 57 of the 74 面 that mean "noodles" were left as 面.
• A wrong but real word. "In the sea", 海里边, contains "nautical mile", 海里(hǎilǐ), so 里 isn't changed to 裡.
• Names. 曹冲(Cáo Chōng), a boy in a well-known story, became 曹衝 instead of 曹沖. In our own tests, the surname 余(Yú) became 餘, "surplus".

Try it on five short paragraphs:

The lesson for learners isn't to avoid converters. It's to know which characters to double-check: the 96 merged ones, and a dozen more that Taiwan keeps apart, such as 游(yóu), which Taiwan writes 游 for "swim" and 遊 for "travel". Our flipper underlines them.

Where is each script used?

Simplified is the official standard in mainland China and Singapore, and Malaysia's Chinese-medium schools have taught it since 1983. Traditional is the standard in Taiwan, Hong Kong and Macau. Everywhere else, it depends on the community, and most large communities use both.

Pick a place to see its script, its Chinese population and the sources:

Sources: China's language law; Taiwan Ministry of Education; Hong Kong Education Bureau; Singapore lists, West 2009; Malaysia, Chinese Wikipedia citing 《语文建设》1982; censuses: Singapore, Malaysia, US, Canada, England and Wales, Australia; Indonesia and the Philippines via Wikipedia. All checked 6 October 2026.

Malaysia is the clearest mixed case. It published a list of simplified characters identical to China's in 1981, yet several Chinese-language dailies print Traditional headlines over Simplified text, according to Chinese Wikipedia, citing a 1982 journal report.

In the United States, weekend Chinese schools belong to two national networks, both founded in 1994. One was started by immigrants from mainland China, and the other mainly by immigrants from Taiwan, says the Center for Applied Linguistics. If you're choosing a school for a child, ask which script it teaches.

Traditional isn't a single standard, either. Taiwan's Ministry of Education fixed its forms in 1982, and Hong Kong compiled its own school reference in 1986.

That reference, Hong Kong's list of standard forms for common characters, 常用字字形表(chángyòngzì zìxíngbiǎo), was re-typeset in 2007, says the Education Bureau.

Among the 1,000 most frequent characters, 9 have different default forms in the two places, by our count. "To speak", shuō, is written 說 or 説, and "inside", lǐ, is 裡 or 裏.

A short history of simplification, 1935 to 2026

China made simplification law in 1956, but the idea is older. The Republic of China published, then halted, its own short list in the 1930s:

Why simplify? The aim was mass literacy in a country where most adults couldn't read, as Wikipedia's history summarises it.

How much the new shapes helped, as opposed to schooling, is still argued. Taiwan and Hong Kong reached high literacy with Traditional characters, as the debate over the two scripts points out.

The 1977 second round shows the limits. Its forms were mostly new inventions rather than shapes people already wrote, and the People's Daily used them for about seven months before dropping them.

Since 1986, the official line has been stability. Traditional characters are allowed in China for set purposes, such as historic sites, calligraphy and hand-written shop signs. Simplified is the standard everywhere else, under Article 19 of the language law.

Taiwan, Hong Kong and Macau never switched

None of them was governed from Beijing when the reforms happened.

The Republic of China had tried, then halted, its own short list in the 1930s. After its government moved to Taiwan in 1949, it kept the characters it already used.

Hong Kong and Macau were under British and Portuguese administration until 1997 and 1999, as Hong Kong's Basic Law and the Macau handover record.

In Taiwan, the script became a political line. A 1993 article in Taiwan Panorama, a government-funded magazine, quoted a Taipei professor saying "simplified characters are like the Red Guards".

The same article reported that a cross-strait agreement signed that spring was printed in both scripts. Taiwan standardised its own character forms in 1982.

Hong Kong's Basic Law makes Chinese and English official languages without naming a script. China's national laws apply in Hong Kong only if they're listed in its Annex III, and the national language law isn't on that list. Hong Kong's teachers, parents and editors work from the Education Bureau's own list of standard Traditional forms.

Which should you learn?

Learn the script of the place and the people you'll read with. If there's no such place yet, start with Simplified. It's the script of the HSK and of mainland China, and adding Traditional reading later is a smaller job than it looks.

Switching costs less than it looks

About a third of the 1,000 most common characters look different. But many of those changes come from shared components, so they arrive in batches.

The hard part runs one way. Going from Traditional to Simplified, merged characters collapse: "to send out", 發(fā), and "hair", 髮(fà), both become 发, and there's nothing to choose.

Going from Simplified to Traditional, you have to learn which form each word takes: the 96 merged groups and a few Taiwan-only distinctions.

Olle Linge of Hacking Chinese estimates "around five hundred tricky cases". He advises waiting until you know a few thousand characters before adding the second set.

Traditional takes more strokes, not more of everything

A Traditional character takes about a quarter more strokes in everyday text, as the stroke chart above shows. Most of what makes Chinese hard is the same in both scripts: vocabulary, tones, and the first 2,000 to 3,000 characters.

Learning Traditional with SayMei

Since 4 October 2026, you can switch SayMei to Traditional characters. The setting is called Chinese characters in Settings. It's called Characters in a story's reading settings, the gear, and in lesson settings, flashcard settings and the Lightning Links game.

Visitors whose connection or browser suggests Taiwan, Hong Kong or Macau are asked once which script they want, as the SayMei changelog notes.

What changes is the Chinese text on screen, including stories, lesson transcripts and the whiteboard, flashcards and Lightning Links. The Traditional worksheet and the level test have their own Traditional options.

Where SayMei falls short, plainly:

• Characters, not vocabulary. You'll see 出租車, not the Taiwan word 計程車.
• The converter isn't perfect. Our audit found 273 wrong characters across our stories, and we're fixing the patterns it found.
• Mainland Mandarin only. Mei Lin's voice and the lessons follow the mainland standard and the HSK. There's no Zhuyin, also called bopomofo, and no Cantonese.
• Tools and dictionary pages stay as written, so a dictionary page shows both forms rather than switching.

What to do next

• Pick the script of your place. Learn the one used by the place and the people you'll read with.
• Default to Simplified if you're unsure. It's the script of the HSK and of mainland China.
• Add Traditional reading later. Learn the shared components first, and whole groups of characters switch at once.
• Double-check the merged characters. Converters and learners slip on the same few, like 里(lǐ), 面(miàn) and 干(gān).
• Read in your target script. Switch a story to Traditional and see how much you can already read.

Try a story in Traditional. Free, no account: open any book, tap the gear, choose Traditional. Open the stories

How we measured

Which characters differ. We ranked characters by frequency in SUBTLEX-CH, a corpus of 46.8 million characters of film and TV subtitles, from Cai and Brysbaert 2010. Then we compared each character with two standards:

1. China's 2013 correspondence table, Attachment 1 of the Table of General Standard Chinese Characters, from the table text.
2. The Taiwan standard forms, from OpenCC's character dictionary and Taiwan variant list.

A character "differs" if any of its Traditional forms differs from it, and "always differs" if it never stays the same. We cross-checked the results four ways:

The 57 Taiwan-only differences are 1955 variant choices such as 于/於, print-shape differences such as 没/沒, and a few rare alternatives OpenCC lists, such as 它/牠. The one China-only case is 才/纔.

Real text. We converted every line of SayMei's 166 graded stories, 344,086 characters at HSK 1–4, with the converter the app uses: opencc-js 1.4.2, Simplified to Taiwan standard, a sentence block at a time. Then we counted the characters that changed.

Converter accuracy. We checked the converted stories four ways, with these results:

The labels are ours and haven't been independently checked yet.

Strokes come from Unicode's Unihan database, kTotalStrokes. HSK lists come from the new HSK syllabus.

Limits. Frequency lists reflect their corpora: subtitles over-represent conversation. "Taiwan forms" follows OpenCC's variant list, which may differ from a given Taiwan textbook in a handful of characters. Population figures come from different years and count different things, like ancestry, ethnicity or citizenship, so compare them with care.