Mandarin über alles

« previous post | next post »

China’s Language Police
Why Beijing Seeks to Extend the Hegemony of Mandarin
By Gina Anne Tam, Foreign Affairs
September 19, 2023

It's odd that the author knows about "topolect" and recognizes the inadequacy of "dialect" as a rendering of fāngyán 方言, but is unwilling to mention "topolect" in this article, which is so suitable for it.  Maybe the unwillingness to shake off that millennial misconception about there only being one Chinese language and a host of "dialects" is part of the problem for the precarious situation in which they find themselves.

In late August, authorities in Hong Kong raided the home of Andrew Chan, the founder of a Cantonese-language advocacy group called the Hong Kong Language Learning Association. National security police questioned Chan about an essay contest the group hosted three years earlier for literature composed in Cantonese, the lingua franca of Hong Kong. One of the finalists in the contest was a fictional futuristic short story about a young man seeking to recover histories of Hong Kong lost to authoritarian erasure. During a warrantless search of his home, they demanded that Chan remove the work from his website, threatening severe consequences for him and his family. Afterward, Chan put out a statement that he had no choice but to dissolve his group entirely, an organization that had worked to promote Hong Kong’s culture through the preservation of the Cantonese language for nearly ten years.

The Chinese state has long been interested in suppressing the diversity of languages spoken in the mainland and, more recently, its special administrative regions. Through state policy, it elevates Mandarin as the sole national language and devalues all other languages, from those spoken by China’s minority ethnicities, such as Tibetan and Uyghur, to other local Chinese languages, the most well-known one being Cantonese. As I observed in Foreign Affairs in 2016, the state language policies that produce this hierarchy are undergirded by the philosophy that Chinese identity, including the language that represents it, should be unified, homogenous, and intrinsically tied to the Chinese state. Such a philosophy sees expressions of Chinese identity that are different or diverse—including the celebration or equal treatment of any language besides the Chinese national language—as unimportant at best, and threatening at worst. In recent years, however, the Chinese state has taken an even more uncompromising line. Its policies of the mid-2010s seem mild by comparison to its attempts today to flatten the complexity of Chinese identity and extend the untrammeled hegemony of Mandarin.


It is common to think of China as a linguistically homogenous country whose citizens speak a sole “Chinese” language, Mandarin. But China is an extremely linguistically diverse country. Beyond the dozens of languages spoken by China’s indigenous minority groups, such as Mongolian, Uyghur, and Tibetan, the country is also home to dozens of Chinese languages such as Cantonese, Shanghainese, and Sichuanese. Today, the state calls these Chinese languages fangyan, a term that is almost universally translated as “dialect.” In official rhetoric, state policies, and even the constitution of the People’s Republic of China, the state deems Mandarin its sole national language and the “common language of the Han people.” The hierarchy embedded in these policies serves the interests of the current Chinese state, but it predates the founding of the PRC. Since the Republican period (1911–49), various Chinese states have promoted Mandarin—a standardized language based on the language of Beijing and its surrounding region—as the sole national idiom, and state and nonstate actors have sought to reframe other Chinese languages as nothing more than dialects. And at the level of policy, the Nationalist government, like its Communist successor, promoted a singular Chinese language with policies that were similar to those the PRC would subsequently implement.

The difference between Mandarin and the "dialects" is far more radical than Tam admits.  In one of my next posts, "Voice-activated lights", I will demonstrate that even Mandarin is fraught with internal mutual unintelligibility.

Under the Chinese leader Xi Jinping, the government has taken a further interest in promulgating a common tongue. A 2012 directive empowered state actors to promote Mandarin through the “supervision and inspection” of the language used in both public and private spaces. From yearly “promote Mandarin week” events in local schools, where schoolchildren happily declare “Speak Mandarin, build the China dream together,” to provincial governments banning the use of local languages in administrative offices, to senior party leaders admonishing cadres whose Mandarin is “deficient,” government actors at all levels have taken this directive to heart in big and small ways. Eleven years later, the effects of this policy are increasingly obvious. Surveys across China increasingly show that the number of people who can speak local languages other than Mandarin is dwindling quickly.

For languages spoken by people who are not Han, China’s ethnic majority, the situation is even more dire. In Xinjiang, Uyghurs are habitually detained or punished for speaking their native language, while the propaganda forced on detainees in “reeducation” camps includes education in the Mandarin language. In Tibet, the state has made it harder for people to learn the local language, even arresting a Tibetan language activist in 2016 for asking the state to honor its constitutional commitment to treat all ethnic languages equally. In 2020, protests in Inner Mongolia against cuts to Mongolian language instruction in schools were met with harsh crackdowns and arrests.

Such crackdowns have been generally less harsh for advocates of languages such as Cantonese—a non-Mandarin tongue spoken by China’s ethnic majority—but these languages, too, endure state suppression. The case of the Hong Kong Language Learning Association is the most recent and public, but it comes on the heels of years of heightened rhetoric from both the Chinese central government and its allies in Hong Kong that downplays non-Mandarin Chinese languages as “nothing more than dialects” that do not deserve the kind of status and clout afforded to the Chinese national language.


These crackdowns are, on their own, a portent of the ways in which the Chinese state seeks to extend the hegemony of Mandarin. Yet they represent only a small part of how the state has sought to force China’s languages into a clear hierarchy. For every quashed demonstration or shuttered advocacy group, there are hundreds of decisions being made in Beijing that create new hurdles for learning, speaking, or creating in languages besides Mandarin.

One of the areas in which these hurdles are most obvious is in infrastructural and educational priorities. Today, in mainland China, all romanization of Chinese characters used on street signs, in books, and in public squares are required to be in Hanyu Pinyin, a romanization system based on Mandarin pronunciation; Chinese children are required to learn it in schools, while romanization systems for other pronunciations, which exist only for a tiny percentage of other character-based Chinese languages, are not. Hanyu Pinyin is regularly taught even in Hong Kong schools, while Cantonese romanization systems are rarely taught, promoted, or used. Even censorship apparatuses reflect the state’s linguistic priorities. In 2019, a popular video-based social media app called Douyin (the Chinese version of TikTok) began sending warning messages to users who posted videos in Cantonese to “Please Use Mandarin.” When pressed about the messages, Douyin’s owner, Bytedance, responded that its aim was not to ban the use of Cantonese; it simply lacked the infrastructure to moderate content in Cantonese. The company that had grown in popularity seemingly overnight saw no value in hiring enough Cantonese speakers to ensure that Cantonese posts properly adhered to China’s censorship regulations.  

Ultimately, people who want to speak in or create in their mother tongues can find workarounds to these problems. Speakers of Cantonese or other non-Mandarin languages can use platforms besides Douyin, at least for now. Although Cantonese speakers may learn Pinyin, they can choose, if they wish, to learn Cantonese romanization through other channels or ignore romanization systems altogether. But the state is choosing to invest in some linguistic infrastructures, such as censors fluent in Mandarin, Mandarin romanization systems, or Mandarin education, and not invest in others. Such decisions are often just as insidious as outright bans, as they ensure that fewer people have the means or will to continue to speak languages that do not benefit from state support or public infrastructure. This kind of passive divestment also comes with the veil of plausible deniability. The state can more easily deny that the neglect of non-Mandarin languages amounts to active oppression.


Through its rhetoric, policies, and enforcement priorities, the government in Beijing has made it abundantly clear that it is intent on stifling not only dissent but also alternative ways of expressing Chinese identity. Indeed, when it comes to Hong Kong, Beijing’s government and its allies in the former British colony see attempts to express a unique sense of identity as tantamount to dissent. From repeatedly looking for legal avenues to ban the Cantonese-language protest anthem “Glory to Hong Kong” to consistently adjusting social studies and history curricula to impress upon schoolchildren that they are not “Hong Kongers” but Chinese citizens who live in Hong Kong, the Hong Kong government and the authorities in Beijing seek to ensure that the vision of a homogenous Chinese identity extends to Hong Kongers.

In recent years, Beijing has acquired better means to enforce its goals. Technological advances have granted the Chinese state new ways to surveil and control private spaces. The intimacy with which the government can today shape everyday life in China has further curtailed what freedom people have to express themselves. The widespread use of surveillance, extrajudicial detention, and forced labor in Xinjiang in recent years represents the lengths to which the government will go to extract compliance from the country’s residents. But many of these strategies are quickly becoming widespread throughout the country, in particular since the “zero COVID” measures that began in 2020 normalized active surveillance in everyday life. Although many of these measures are not yet common in Hong Kong, surveillance there, too, has increased, both in how Hong Kong police track an individual’s movements and speech, and in the range of actions they deem threats to national security.

Wake up, Hong Kongers and Taiwanese and Shanghainese and Sichuanese and all you other "dialect" speakers.  If you do not, you will end up with no language of your own.

That does not mean that people in China and Hong Kong are quietly accepting the stiffening of authoritarian rule. Dissent in the mainland, while sometimes furtive and tenuous, can still be seen throughout the country, from efforts by local Shanghainese to resist Mandarin hegemony by promoting their local language through literature contests to the "white paper" protests against the zero-COVID lockdowns. But resistance to Beijing’s increasing authoritarian rule is most palpable in Hong Kong. The 2019 protests brought the unique identity of Hong Kong to the forefront, as more and more Hong Kongers refused to identify with their overarching national identity, with polls at the time finding that fewer than one in ten Hong Kong residents identified as exclusively Chinese, and almost half identified only as “Hong Konger.” This shift is even more noticeable among young people, as a June 2022 poll found that 76 percent of people aged 18 to 30 identified as Hong Kongers, with only two percent describing themselves as Chinese.

Language has been a core vector through which Hong Kongers have resisted state attempts to reshape the identity of their city. Indeed, as journalists such as Mary Hui have noted, Cantonese became a core language of protest in the 2019 movement, a way for Hong Kongers to assert their identity as separate from that of the PRC and create a shared set of symbols, phrases, and songs that bonded Cantonese speakers together in the context of opposition to the PRC state. In the wake of the 2020 national security law, a sweeping law that nominally targets acts of secession and subversion but is being used liberally to squash dissent and protest, efforts to protect Hong Kong identity through the preservation of its language have extended around the world, with Cantonese-speaking groups in North America and Europe working to promote the language within the diaspora.

The effect of the closure of the Hong Kong Language Learning Association is clear: Cantonese, and non-Mandarin languages in China writ large, have lost a strong advocate, and others who seek to promote language rights will become wary of doing so. And the effects of this chilling of speech are wide-reaching. Language is an integral part of who people are. When a powerful institution narrows where and how they can use it, as the Chinese state has been doing in the past decade, it stifles their ability to express themselves in complex, deeply human ways.

It would also help your case if you would develop phonetic writing systems that adequately represent your languages instead of Chinese characters, which work well for Literary Sinitic / Classical Chinese, but do not even do a good job for spoken Mandarin topolects, witness the complaints of the great writer, Lao She (1899-1966).

Two things I [VHM] never do:  refer to Sichuanese, Cantonese, Taiwanese, etc. as "dialects" or "slang" (the latter is especially demeaning).  I simply call them "topolects", which is a neutral designation for them and which is also an accurate translation of the Chinese term fāngyán 方言.

"The American Heritage Dictionary of the English Language, 5th edition"   (11/14/12)

"Mutual intelligibility" (5/28/14)
(see the long list of posts linked at the bottom)

"What Is a Chinese “Dialect/Topolect”? Reflections on Some Key Sino-English Linguistic Terms," Sino-Platonic Papers, 29 (1991).

Also here and, for a complete list of my LL posts dealing with topolects, see here.

See, as well, "The Classification of Sinitic Languages: What is 'Chinese'", which is a chapter from this book:

Breaking Down the Barriers: Interdisciplinary Studies in Chinese Linguistics and Beyond


Selected readings

[h.t. Norman Leung and Don Keyser]


  1. Jonathan Smith said,

    September 19, 2023 @ 6:57 pm

    As you say yourself, topolect is a neutral (indeed, democratic, egalitarian…) term. Of course it is a poor way to translate fangyan, which is (and has long been) the choice of the state exactly because it is stigmatizing — that is, has none of the above listed qualities. The compositional parallelism is surely irrelevant? Would you really have it that Mandarin über alles voices referred to non-Mandarin languages as "nothing more than topolects"?

  2. Victor Mair said,

    September 19, 2023 @ 7:06 pm

    I can't change the Chinese, but I can deal with our misunderstanding of the Chinese, which only compounds the deficiencies of their usage.

  3. Jenny Chu said,

    September 19, 2023 @ 9:20 pm

    I think it's not surprising that in Foreign Affairs magazine, which is written for a general audience, the author should choose "languages" as a term that doesn't need to be explained, rather than "topolects", which does. Don't forget that the default assumption of many readers going in is that there is, indeed, just one "Chinese" language which is spoken by all Chinese people.

    Anyway – the whole "we don't have enough moderators to manage Cantonese content" is, indeed, a sack of … nonsense. Imagine if Facebook said, "Oh, we can't moderate content in Italian – please use English!" (There are fewer Italian speakers on this planet than there are Cantonese speakers, by the way.)

  4. Victor Mair said,

    September 19, 2023 @ 10:12 pm

    I'm talking about her use of "dialects" instead of "topolects". She needs to clarify what she means by "dialects" in contrast to what the Chinese mean by "topolects" (fāngyán 方言).

    I'm all in favor of "languages" for Cantonese, Taiwanese, etc.

  5. Philip Spaelt said,

    September 19, 2023 @ 11:43 pm

    @VM: I don't understand your complaint. The author uses "dialect" 3 times in the article always in an attributive sense, as "what others call them".

  6. Tom said,

    September 20, 2023 @ 1:52 am

    @Victor, I don't understand your take. Tam is mentioning other people's use of the term "dialect" in order to criticize it. And other people *do* refer to non-Mandarin Sinitic languages as "dialects," in spite of the situation. Are you saying that Tam shouldn't mention this phenomenon at all, lest the label get inadvertently perpetuated?

  7. Weh said,

    September 20, 2023 @ 2:55 am

    The Chinese term 方言 is in fact used almost identically to the English "dialect", e.g. when describing the dialects of English or Japanese (which also uses the term 方言 "hōgen"). On the other hand, the major Chinese languages are traditionally referred to as "languages" 语 (粤语、吴语、闽南语, etc).

    In "The Chinese Journal of Language Policy and Planning" (《语言战略研究》2018年第3期) there is an article written by 李佳 and titled «"Yut language" (~Cantonese), "Banlam language" (~Hokkien) and "Whu language" (~Shanghainese): three cases of referring to Chinese dialects as "languages"» (《“粤语”“闽南语”和“沪语”:汉语方言称“语”的三种形成模式》), in which he provides some examples of interchangeability between the words "language" and "dialect" even in the early 20th century. For instance, the Wuhan University was known in 1902-1922 as 方言學堂, which should be translated to "The Foreign Languages Academy". He also points out the usage of the word "language" 语 in local documents and media, where terms like 闽南语 are more frequent than 闽南话, and much more common than 闽南方言. Sometimes the regional languages are even distinguished from the 汉语 "Han (or Chinese) language", and 李佳 provides an example from an article by Xinhua News Agency:

    «We can sing "Happy Birthday" in 5 languages: Chinese, English, Uyghur, Kazakh, and Cantonese.»

    As a pro-party linguist, however, 李佳 considers this usage detrimental to the national unity and ideology, and calls for more regulations to prevent govenment agencies and media from using the word "language" 语 for what he considers "dialects" 方言.

  8. Lasius said,

    September 20, 2023 @ 3:07 am

    Don't know what this has to do with the phrase "über alles" from the Deutschlandlied? That line is talking about an ideal of German unity as opposed to fragmentation into minor states, not about dialects or languages.

    Wouldn't France's language policies be a much closer fit?

  9. Cultrev said,

    September 20, 2023 @ 6:25 am

    It is always a good reminder or an alert for those who would have fantasy for the CCP or the country ruled by it – how the great writer, Lao She (1899-1966) ended up his life.

  10. Jonathan Smith said,

    September 20, 2023 @ 8:27 am

    To be precise, fangyan and cognates aren't really parallel to topolect even etymologically — fang is not simply 'place' but '(outlying) region'; this is why fangyan could mean 'foreign language' in the early 20th century as noted by Weh.

    IMO "粤语"、"吴语"、"闽南语" etc. represent a relatively recent (relatively) enlightened turn (plus other details… e.g., I am not sure if the second in particular is really used to refer to the local language(s) by native speakers or remains just a technical classificatory term, while Banlamgi is seen as politically problematic as opposed to enlightened by many Taiwanese.) The experience/memory of these communities is instead largely of pejorative reference to their speech as "only" (Tw.) hōgen 方言, etc., by would-be language authorities.

  11. Victor Mair said,

    September 20, 2023 @ 9:11 am

    First sentence of the first comment above: "…topolect is a neutral (indeed, democratic, egalitarian…) term".

    Confusion reigns over "dialect" because few, if any, are willing to confront it head on. It is not identical with "topolect" (fāngyán 方言). It merits a thorough critique and debunking, especially from specialists who use it, whether attributively or not, without explaining how it differs from fāngyán 方言 ("topolect"), yǔyán 語言 ("language"), and other relevant terms.

    Among pro-Party linguists, there is an intense fear of the fragmentation of the imposed, artificial national language under threat from the natural, local lects and languages. Hence, Lǐ Jiā 李佳 and others of his ilk call "for more regulations to prevent govenment [sic] agencies and media from using the word 'language' 语 for what he considers 'dialects' 方言" — except that they don't really call them "dialects", they call them fāngyán 方言 ("topolects"), whatever they mean by that.

    Cultrev is right on. Lao She loved Pekingese, but he was not permitted to use it in all its glory because of the Mandarin-centric script and Confucian-Communist ideology.

  12. Jerry Packard said,

    September 20, 2023 @ 9:25 am

    The promotion of the standard dialect has been a right-left issue since before Mao, with the rightists wanting to promulgate the standard dialect and the leftists pushing for local dialects.

    I agree with Jenny Chu that ‘dialect’ is better used here as a non-technical term for general readers.

    Regarding the translation of 方言, I have always preferred John DeFrancis’ term ‘regionalect’, because it is easily understood, and because it has a smooth derivation: region > regional > regionalect.

  13. Victor Mair said,

    September 20, 2023 @ 9:47 am

    "regionalect" (3,830 ghits)

    "topolect" (28,300 ghits) — has been in the last two editions of the American Heritage Dictionary of the English Language


    In terms of size and space / location, consider how these expressions are used by linguists and in casual speech:

    次方言 ("subdialect/topolect") 104,000 ghits
    小方言 ("small dialect/topolect") 133,000 ghits

  14. Michael Watts said,

    September 21, 2023 @ 4:20 am

    Imagine if Facebook said, "Oh, we can't moderate content in Italian – please use English!"

    But Facebook moderation is primarily conceived of as being for the benefit of the users. If Facebook's goals were to suppress certain viewpoints commonly held by Italians, and they didn't feel they could trust an Italian speaker not to sympathize with those viewpoints, they would quickly find that they were indeed unable to moderate Italian-language content.

  15. Victor Mair said,

    September 22, 2023 @ 5:27 am

    "because it is easily understood, and because it has a smooth derivation"

    That's "topolect".

  16. Victor Mair said,

    September 27, 2023 @ 7:08 am

    ABC (Australian Broadcasting Corporation) radio program (9/26/23):

    "One country, one tongue: why China is suppressing language diversity", Guest: Gina Anne Tam

  17. KIRINPUTRA said,

    September 28, 2023 @ 6:47 am

    Having read through the comments, I see what @Victor is saying. But why translate a non-working concept?

    Among other things, the modern Mandarin (or Neo-Chinese; "中文") word 方言 assumes that each locale has one local Sprache. Even ignoring the overlay of Mandarin, this assumption doesn't fit reality in tropical & near-tropical China.

    For example, the Baidu article about Luichew City 雷州市 ( implies that the city is mostly Luichew-speaking 雷州話; Luichew is THE lect of the locale. It ignores the dominance of (LOCAL) Cantonese over all other non-Mandarin languages. This "oversight" is built into the broken idea of 方言. The idea of a micro-locale being home to two or more linguistic groups that geographically overlap is literally unacceptable to somebody thinking in terms of 方言.

    To give (the modern word) 方言 more of the benefit of the doubt, I wonder if it might not just generally be an approximation of what English speakers (in the Philippines, for instance) would call "dialect" anyway. There is almost invariably no intent to give the 方言 any kind of meaningful respect by calling it that.

    Also — and this is beyond my literacy & experience — doesn't the term 方言 itself imply a Sprache that lacks "vertical integration" with a homegrown, well-fitting writing system?

RSS feed for comments on this post