Are Chinese topolects dying?

Two things are true at once and people keep picking one. No major Chinese variety is classified as endangered by the international databases. And in the cities, the number of children growing up speaking one is falling fast.

Updated 22 September 2026 · 5 min read

The number that explains everything else

In 2020 the Ministry of Education reported that 80.72 per cent of China’s population spoke Putonghua — up 27.66 percentage points from 2000, which puts the starting point at about 53 per cent.

Read that again slowly. In twenty years, roughly a quarter of the population of China acquired a language. There is no comparable peacetime language shift anywhere, and it was policy working exactly as designed. Everything else on this page is a consequence of it.

What the law actually says

The framework is the Law on the Standard Spoken and Written Chinese Language, passed on 31 October 2000 and in force from the start of 2001. Article 2 defines the standard as Putonghua, with pronunciation based on the Beijing dialect, plus the standardised characters. Article 3 says the State popularises them.

The part rarely quoted is Article 16, which is where the other varieties live. It permits local speech in defined circumstances: where state functionaries genuinely need it for their duties, in broadcasting with the approval of the provincial or national broadcasting authority, and where it is needed in traditional opera, film, television and other arts.

So this is not a ban. It is a law that makes one variety the default everywhere and makes the others require a reason and a permission. That distinction matters for predicting what happens: languages rarely die of prohibition, and frequently fade when every institutional default points the other way.

The distinction nearly every article gets wrong

Here is the crux. A branch with eighty million speakers is not endangered. A city where children no longer learn it has a transmission problem. Both statements can describe the same language, and confusing them produces either complacency or alarmism depending on which you pick up.

The evidence for the first is straightforward: Wu, Yue, Min, Hakka, Xiang, Gan and Jin do not appear in the list of endangered languages in China drawn from UNESCO’s atlas. The entries there are minority and non-Sinitic languages with far fewer speakers.

And the disagreement is on the record. Wikipedia’s account of Wu puts it precisely: the Chinese dialectologist Cao Zhiyun “considers many of these languages ‘endangered’ and has introduced the term 濒危方言 (‘languages in danger’) to raise people’s attention to the issue, although major international databases, such as Glottolog and Ethnologue, do not share similar sentiments”.

Neither side is being careless. Cao is looking at who is speaking it to children in Shanghai; the databases are counting speakers. Those measurements genuinely diverge, and they diverge most in exactly the cities where these languages are most famous.

Shanghai: the transmission case

Shanghai is the clearest example of a large language with a small-language problem. Shanghainese use was discouraged in schools from 1992 — students were banned from speaking it on campus — and the effect on the generation raised afterwards is visible. Around forty per cent of Shanghai residents under eighteen cannot speak it fluently.

The full picture, with the speaker numbers, the surveys and what is being done, is on the page about Shanghainese specifically. The point here is that Wu as a branch is in no danger whatever while Wu in its most prestigious city is not being passed on.

Cantonese: the case that is less bad than you have read

Cantonese is the variety most often described as threatened, and the hard data from Hong Kong does not support the stronger versions of that claim. The 2021 census found that the share of the population aged five and over able to speak Cantonese went from 97.4 per cent in 2011 to 96.0 per cent in 2021, and as a usual spoken language from 91.5 to 90.6 per cent.

That is a marginal decline against a decade of enormous social change, alongside rising ability in Putonghua — from 49.5 to 56.5 per cent — mostly as a second language rather than as a replacement. What Hong Kong shows is bilingualism increasing, not Cantonese collapsing.

Guangzhou is a different story, and the more worrying one, because it is a mainland city under mainland policy. And it produced the clearest public test of how people feel about it: in July 2010 a proposal to increase Mandarin programming on Guangzhou Television drew opposition from 79.5 per cent of thirty thousand people surveyed, then street protests, then more than twenty arrests. The proposal was dropped.

Taiwan: the one place that reversed

For most of the twentieth century Taiwan did to Hokkien what the mainland did to everything else, and more harshly: restricted in schools and broadcasting from 1964, with violations at school “often resulting in physical punishments, fines, or humiliation”. Then it turned around.

Martial law was lifted in 1987, and the mother-tongue movement followed in the 1990s.
In 1993 Taiwan became the first country in the world to teach Taiwanese Hokkien in schools; by 2001 Hokkien, Hakka and indigenous languages were taught in all of them.
The Development of National Languages Act took effect in 2019, making them national languages, with mandated promotion and mother-tongue instruction through secondary school.
In September 2025 Taiwan opened its first Taiwanese-language immersion primary school.

And here is the sobering part, which anyone hoping policy alone will fix this should sit with: despite all of that, census data shows the number of people speaking Taiwanese has continued to drop. Thirty years of supportive policy, official status, school instruction and dedicated broadcasting have slowed a decline rather than reversed it.

So: are they dying?

The honest answer has three parts.

Not as branches. Wu, Yue, Min and the rest have tens of millions of speakers each and no international body classifies them as endangered.
Yes, in the cities, among children. That is where a language’s future is decided, and in Shanghai and Guangzhou the trend is clear and not new.
Slowly, and less predictably than either side claims. Predictions of language death have a poor record. The specific figure you will see quoted for Shanghainese — that it will be gone within fifty years — comes from a preservationist rather than from academic work, and should be read as a warning rather than a forecast.

What actually determines the outcome is not broadcasting minutes or street protests but what parents speak at the dinner table, which no policy reaches directly. That is also the most hopeful thing about it: the mechanism is entirely in ordinary people’s hands, and it costs nothing.

The part that is genuinely looking up

Popular interest has moved in the other direction lately. Wong Kar-wai’s 2023 series Blossoms Shanghai, released in a Shanghainese version as well as a Mandarin one, is widely credited with reviving attention to regional speech, and it ran on a national channel — which Article 16 permits precisely because drama is one of the listed exceptions. Popular support for preserving Wu is described as very strong.

Whether attention becomes speakers is the open question, and nobody has measured it. That series has its own page here, and the broader family it belongs to is on the Wu Chinese page.

Are Chinese topolects dying?

The number that explains everything else

What the law actually says

The distinction nearly every article gets wrong

Shanghai: the transmission case

Cantonese: the case that is less bad than you have read

Taiwan: the one place that reversed

So: are they dying?

The part that is genuinely looking up

Sources

Learn Shanghainese with Verbavia

Read next

The Shanghainese course and word list