« previous post | next post »

For May 21, China Real Time Report, the China blog of the Wall Street Journal, featured an article entitled "Do You Dare Try the Devil-Language? China’s 10 Hardest Dialects" by Isabella Steger.

Here's the full list of "hardest dialects", with many obvious omissions, such as Fuzhou and Chaozhou / Teochew:

1. Wenzhou dialect

2. Cantonese

3. Suzhou and Minnan dialects

4. Shanghainese

5. Shaanxi dialect

6.  Changsha dialect

7. Sichuan dialect

8. Shandong dialect

9. Tianjin dialect

10. Northeast dialect

This is quite a motley collection of branches, languages, and topolects, most of which are erroneously called such-and-such a dialect, and two of which are referred to as "-ese".  Suzhouese and Minnan (no. 3) belong to completely different branches of Sinitic, while Wenzhou, Tianjin, Changsha, and Shanghai are cities, Shandong, Shaanxi, and Sichuan are provinces, and the Northeast is a region (used to be called Manchuria).  Half of the items (5, 7, 8, 9, and 10) are considered to be types of Mandarin, though the degree of intelligibility among them varies greatly; the amount of intelligibility among the remaining five, and between those five and the five supposedly Mandarin types is zero or close to it.  The five non-Mandarin types belong to or constitute separate, non-Mandarin branches of Sinitic.

One of the commenters (Zhang) astutely notes that, if you are referring only to mutually unintelligible spoken varieties, "then there will be thousands of languages in China."

The Chinese term for "devil-language" is guǐhuà 鬼话.  Here it refers to the kind of language spoken by the devil, but in Mandarin it commonly has the extended meaning of "lie; nonsense; bullshit; deception; humbug".

We have previously come across the notion that one of the non-standard varieties of Sinitic is associated with the devil.  See "Is Cantonese a language, or a personification of the devil?"

Two days later, on May 23, China Real Time Report followed up a special report on the number one allegedly "hardest dialect" in China, "What It’s Like to Live in China and Speak the 'Devil-Language'", by Lillian Lin.

The author states that, when she was living in a college dorm in Beijing, none of her roommates, "who were from five cities across China, could understand a single word of [her] conversations."

If you want to hear what Wenzhou language sounds like, there's a half-hour video with ample samples at the end of this article.

Wenzhouese is generally thought of as a highly divergent member of the Wu branch of Sinitic (includes the languages of Suzhou, Hangzhou, Shanghai, etc.).  In actuality, Wenzhouese should be considered as a group of languages, since several of its constituent varieties have a very low degree of intelligibility among them, as observed by commenter Harland:

Wenzhou dialect isn’t just one monolithic thing, as this article suggests. It’s a whole family of languages in itself. Try speaking Wenzhou language in Ruian or Longyan and see how far you get. It’s a Tower of Babel out here.

If all of this talk about "devil language" and mutually unintelligible varieties of Sinitic teaches us anything, it is that "Chinese" is not a single, monolithic tongue and thousands of dialects that are all the same when written down (one of the grossest myths that is accepted by gospel truth by most people who know nothing about the real situation), but that it consists of innumerable languages and many branches that remain to be classified in a systematic fashion comparable to what we have for Indic, Romance, Germanic, Slavic, and so forth.

Here follows a sampling of the many Language Log posts dealing with the questions of intelligibility, dialect, and language raised above:

"Linguistic diversity in Greater Tibet"

"Spoken Hong Kong Cantonese and written Cantonese"

"English is a Dialect of Germanic; or, The Traitors to Our Common Heritage"

"Zazaki: a West Iranian language"

"Uyghur as a 'dialect' — NOT"

"Mutual Intelligibility of Sinitic Languages

"The American Heritage Dictionary of the English Language, 5th edition"

"Shandong Dialect Intelligibility"

Based on the materials presented and referred to in the above and other Language Log posts, we need to keep the following caveats in mind when discussing questions of dialect and language in the Chinese context:

1. There is a huge gulf between spoken and written language in China.

2. There are countless varieties of spoken Sinitic, many of which are mutually unintelligible.

3. An adequate taxonomy and classification of the Sinitic family have yet to be adumbrated.

4. "Dialect" is at best a misleading English translation of fāngyán 方言; "topolect" is far more accurate.

5. A precise Chinese translation of "dialect" would be something like xiāngyán 相言.

6. Throughout history, there have basically been two different types of written Chinese:  Literary Sinitic and Vernacular Sinitic (a koine or lingua franca).  Since around 1920, China has had a written national language that we may refer to as Modern Standard Mandarin (MSM).

7. The non-standard vernaculars (i.e., all of the non-MSM vernaculars) are almost never written down, and when on rare occasions they do appear in written form, it is usually only as bits and pieces embedded in a fundamentally Mandarin matrix or as a sort of tour de force requiring numerous special characters or in romanized or other phonetic transcription.

8. Reading aloud a written Chinese text in the pronunciations of the various topolects is a very different matter from the unadulterated forms of the topolects as spoken in daily life.

Please, no talk of armies and navies.

[Hat tip to June Teufel Dreyer.]


  1. Movenon said,

    May 25, 2014 @ 9:24 pm

    Haha I think this has been circulating on the Chinese web for quite a number of years now. Here are my 10 additions to extremely difficult Sinitic tongues:

    1. Fuzhounese – I am currently studying this, and it trumps all other Sinitic varieties I have ever studied in difficulty, including Cantonese, Minnan, Teochew, Shanghainese, Mandarin, Taishanese, Hakka, and even Wenzhounese.

    2. Putian Min – Originated as a basically Minnan dialect that diverged off and became phonologically influenced by the difficult Fuzhounese, taking up many of its peculiar phonological features, as well as developing more of its own. If you slow them down and ask for some isolated words, you can notice the strong Minnan resemblance, but at fast spoken speeds of course it's unintelligible.

    3. Hainanese – as hard as the other Southern Min lects, and went through lots of unique sound changes that make it very different from mainstream Minnan/Teochew.

    4. Teochew – same difficulty level as the closely related Hokkien Minnan; one extra tone but makes up for it by having simpler tone sandhi.

    5. Nantong dialect – intermediate lect between Northern Wu and Jianghuai Mandarin, unintelligible with either one.

    6. any of the Southern Hunan lects (Old Xiang, etc.) – less transparent than the highly Mandarinized Changsha dialect, lots of archaic features like voiced consonants

    7. any of the Huizhou dialects (like Tunxi, etc.) – I've met a couple speakers of Tunxi-hua, and it was very different and unique from other Sinitic lects.

    8. any of the Lishui (in Zhejiang) Wu lects – Qingtian-hua is relatively common in Chinese communities of Europe. Does have similarities with both Shanghainese and Wenzhounese, but not enough for mutual intelligibility with either one. Has a good bit of an archaic Min substratum as well (the word for 'pig' happens to be identical to Taiwanese Minnan).

    9. any Pinghua or Tuhua lect – these lects aren't even on the radar for most Chinese (most Chinese people have at least heard of Wenzhou, Qingtian, Hainan and other places for example). I met a Hunan Tuhua speaker before, it was quite interesting to hear it spoken.

    10. lots of tiny dialect islands scattered in the South – Longdu Min in Zhongshan, Manjiang and Manhua in southern Zhejiang, etc.

    and many more!

  2. Ken said,

    May 25, 2014 @ 10:20 pm

    They can't be that hard, two-year-old children can learn them.

    (Tongue only half in cheek. It seems that "hard" here means "if you grew up within twenty miles of Beijing, you'll have trouble with these"?)

  3. Mark Mandel said,

    May 25, 2014 @ 10:37 pm

    @Ken: Maybe in Steger's report, but I suspect that for Movenon it's more like "if you didn't grow up within thirty kilometers of where they're spoken, you'll have trouble with these".

  4. Movenon said,

    May 25, 2014 @ 10:50 pm

    Oh yes, I'll clarify, I was generally going off "from the starting position of L1 Putonghua" which I believe was the same premise as the original article in Chinese that's been circulating for years. Of course little kids will figure out how to speak whatever form(s) of speech they're predominantly exposed to as a child.

    "X kilometers away from X point" can be very tricky and imprecise when it comes to Sinitic lects, because there are some stark boundaries (like for example between Teochew on one side and Yue/Hakka on the other) and some not-so-clear boundaries, like between Yue and Hakka at certain points, or between Soutwestern Mandarin and Xiang at some points.

  5. Simon P said,

    May 25, 2014 @ 11:55 pm

    Surely nobody dares to speak Cantonese. It gives you cancer:,127452

  6. Vanya said,

    May 26, 2014 @ 3:47 am

    I imagine it is very hard to learn any language that has no written form, is spoken only among a fairly small community of like-minded speakers who share cultural and geographical references of which outsiders are not aware, and lacks a standardized "elite/educated" language that can bridge various regional variations within that language.

  7. Peter said,

    May 26, 2014 @ 5:06 am

    @Ken, @Mark Mandel: surely “hard”, in all these cases, mainly means “lacking helpful similarities to standard Mandarin”. Presumably, for (say) an anglophone with no previous knowledge of Sinitic languages, Wenzhou is not intrinsically harder to learn than Mandarin? (Of course, in practice Mandarin is easier due to better availability of teaching materials, etc.)

  8. Victor Mair said,

    May 26, 2014 @ 6:00 am

    @Simon P

    Many thanks for the great link. Language Log readers are encouraged to look at the comment thread to the Sheik post, where even more ridiculous views are cited. Follow the links. It is encouraging that Wikipedia is not taken in by all of this:

    See this section:

    Here's the amazing abstract of the "scientific" paper ("Is nasopharyngeal cancer really a 'Cantonese cancer'?") cited in the third comment:


    Nasopharyngeal cancer (NPC) is endemic in Southern China, with Guandong province and Hong Kong reporting some of the highest incidences in the world. The journal Science has called it a "Cantonese cancer". We propose that in fact NPC is a cancer that originated in the Bai Yue ("proto Tai Kadai" or "proto Austronesian" or "proto Zhuang") peoples and was transmitted to the Han Chinese in southern China through intermarriage. However, the work by John Ho raised the profile of NPC, and because of the high incidence of NPC in Hong Kong and Guangzhou, NPC became known as a Cantonese cancer. We searched historical articles, articles cited in PubMed, Google, monographs, books and Internet articles relating to genetics of the peoples with high populations of NPC. The migration history of these various peoples was extensively researched, and where possible, their genetic fingerprint identified to corroborate with historical accounts. Genetic and anthropological evidence suggest there are a lot of similarities between the Bai Yue and the aboriginal peoples of Borneo and Northeast India; between Inuit of Greenland, Austronesian Mayalo Polynesians of Southeast Asia and Polynesians of Oceania, suggesting some common ancestry. Genetic studies also suggest the present Cantonese, Minnans and Hakkas are probably an admixture of northern Han and southern Bai Yue. All these populations have a high incidence of NPC. Very early contact between southern Chinese and peoples of East Africa and Arabia can also account for the intermediate incidence of NPC in these regions.


    By the way, the first commenter has this note at the bottom of his posts:

    我係鬼仔 我用倉頡輸入中文

    Translation: I am a "[foreign] devil" (i.e., "foreigner"); I use Cangjie / Tsang-chieh (Cantonese Cong1kit3 / ChongKit) to input Chinese.

    For more information on "gweilo / gwailo" ("[foreign] devil"), follow these links:

    Oh, and don't miss the comments here:

    Lots of funny stuff, and even some good, useful information.

  9. Mark Mandel said,

    May 26, 2014 @ 10:00 am

    @Peter: I was riffing on Ken's "if you grew up within twenty miles of Beijing, you'll have trouble with these".

RSS feed for comments on this post