Chinese (hànyǔ 汉语/漢語 or zhōngguóhuà 中国话/中國話) comprises many regional language varieties sometimes grouped together as the Sinitic languages, the primary ones being Mandarin, Wu, Cantonese, and Min. These are not mutually intelligible, and even many of the regional varieties (especially Min) are themselves composed of a number of non-mutually-intelligible subvarieties. As a result, Western linguists typically refer to these varieties as separate languages. For sociological and political reasons, however, most Chinese speakers and Chinese linguists consider them to be variations of a single Chinese language, and refer to them as dialects, translating the Chinese terms huà 话, yǔ 語, and fāngyán 方言. 

The neologism topolect has been coined as a more literal translation of fangyan in order to avoid the connotations of the term "dialect" (which in its normal English usage suggests mutually intelligible varieties of a single language), and to make a clearer distinction between "major varieties" (separate languages, in Western terminology) and "minor varieties" (dialects of a single language). In this article, however, the generic term "variety" will be used.

Chinese people make a strong distinction between written language (文, Pinyin: wén) and spoken language (语/語 yǔ). English does not necessarily have this distinction. As a result the terms Zhongwen (中文) and Hanyu (汉語/漢語) in Chinese are both translated in English as "Chinese". Within China, it is common perception that these varieties are distinct in their spoken forms only, and that the language, when written, is common across the country.

Classification

Chinese consists of several dialect continuums. Differences in speech generally become more pronounced as distances increase, with few radical breaks. However, the degree of change in intelligibility varies immensely depending on region. For example, the varieties of Mandarin spoken in all three northeastern Chinese provinces are mutually intelligible, but in the small province of Zhejiang a person from one valley may be completely unable to comprehend the language from the next, though both are considered dialects of Wu Chinese.

In the book, "The Middle kingdom: a survey of the ... Chinese empire and its inhabitants ...", published in 1848, the different varieties of Chinese were described as "dialects", the book acknowledged that they were mutually unintelligible and the term "dialect" was used in a different sense than the western term, in which a dialect was merely indicative of a small difference in pronunciation, while in China, the entire grammar and idiom were different, the written language was what united the different Chinese dialects.

Mandarin (Standard Chinese) is the dominant variety, much more widely studied than the rest. Outside of China, the only two varieties commonly presented in formal courses are Mandarin and Cantonese. Inside China, second-language acquisition is generally achieved through immersion in the local language.

The scientific classification of Chinese into different regional dialects is very recent. The first such efforts were made by Fang-kuei Li in 1937, which, with only minor modifications, form the basis for the current, conventionally accepted set of seven dialect groups:

Mandarin 官话/官話 (also Northern 北方話/北方话): (c. 836 million speakers) This is the group of dialects spoken in northern and southwestern China, and makes up the largest spoken language in China. Standard Chinese, called Putonghua or Guoyu in Chinese, which is often also translated as "Mandarin" or simply "Chinese", belongs to this group. It is the official spoken language of the People's Republic of China, and Singapore. Mandarin Chinese is also the official language of the Republic of China governing Taiwan, although there are minor differences in this standard from the form standardized in the PRC.[3]

Wu 吴语/吳語: (c. 77 million) spoken in the provinces of Jiangsu and Zhejiang, and the municipality of Shanghai. Wu includes Shanghai dialect, sometimes taken as the representative of all Wu dialects. Wu's subgroups are extremely diverse, especially in the mountainous regions of Zhejiang and eastern Anhui. The group possibly comprises hundreds of distinct spoken forms which are not mutually intelligible. Wu is notable among Chinese dialects in having kept "voiced" (actually slack voiced) initials, such as /b̥/, /d̥/, /ɡ̊/, /z̥/, /v̥/, /d̥ʑ̊/, /ʑ̊/ etc.

Yue (Cantonese) 粤语/粵語: (c. 71 million) spoken in Guangdong, Guangxi, Hong Kong, Macau, parts of Southeast Asia and by Overseas Chinese with an ancestry tracing back to the Guangdong region. The term "Cantonese" may cover all the Yue dialects, including Taishanese, or specifically the Canton dialect of Guangzhou and Hong Kong. Not all varieties of Yue are mutually intelligible. Yue retains the full complement of Middle Chinese word-final consonants (p, t, k, m, n, ng), and has a well-developed inventory of tones.

The Min languages 闽语/閩語: (c. 60 million) spoken in Fujian, Taiwan, parts of Southeast Asia particularly Malaysia, Philippines, and Singapore, and among Overseas Chinese who trace their roots to Fujian and Taiwan, particularly prevalently in New York City in the United States. The largest Min language is Hokkien, which is spoken in Southern Fujian, Taiwan, and by many Chinese in Southeast Asia and includes the Taiwanese, and Amoy dialects amongst others. Min is the only branch of Chinese that cannot be directly derived from Middle Chinese. It is also the most diverse, divided into seven subgroups defined on the basis of relative mutual intelligibility: Min Nan (which includes Hokkien and Teochew), Min Dong (which includes the Fuzhou dialect), Min Bei, Min Zhong, Pu Xian, Qiong Wen, and Shao Jiang.

Xiang (Hunanese) 湘语/湘語:(c. 36 million) spoken in Hunan. Xiang is usually divided into the "old" and "new" dialects, with the new dialects being significantly influenced by Mandarin.

Hakka 客家话/客家話: (c. 34 million) spoken by the Hakka people, a cultural group of the Han Chinese, in several provinces across southern China, in Taiwan, and in parts of Southeast Asia such as Malaysia and Singapore. The term "Hakka" itself translates as "guest families", and many Hakka people consider themselves to be descended from Song-era and later refugees from North China, although their genetic origin is still disputed. Hakka has kept many features of northern Middle Chinese that have been lost in the North. It also has a full complement of nasal endings, -m -n -ŋ and occlusive endings -p -t -k, maintaining the four categories of tonal types, with splitting in the ping and ru tones, giving six tones. Some dialects of Hakka have seven tones, due to splitting in the qu tone. One of the distinguishing features of Hakka phonology is that Middle Chinese voiced initials are transformed into Hakka voiceless aspirated initials.

Gan 赣语/贛語: (c. 31 million) spoken in Jiangxi. In the past, it was viewed as closely related to Hakka dialects, because of the way Middle Chinese voiced initials have become voiceless aspirated initials, as in Hakka, and were hence called by the umbrella term "Hakka-Gan dialects".

Ba-Shu, of Sichuan, was one of the most divergent varieties of Chinese. However, it was supplanted by Southwestern Mandarin during the Ming dynasty.

There is some dispute as to whether the following varieties should be classified separately:

Huizhou 徽语/徽語: (c. 3.2 million) spoken in the southern parts of Anhui—formerly, and sometimes still, classified as a dialect of Wu, now classified as an independent dialect.

Jin 晋语/晉語: spoken in Shanxi, as well as parts of Shaanxi, Hebei, Henan, and Inner Mongolia. Often classed as dialect of Mandarin.

Pinghua 平话/平話: (c. 2 million) spoken in parts of the Guangxi. Sometimes classed as dialect of Cantonese.

Some varieties remain unclassified. These include:

Danzhou dialect 儋州话/儋州話: spoken in Danzhou, Hainan.

Xianghua 乡话/鄉話: spoken in a small strip of land in western Hunan, this group of dialects has not been conclusively classified.

Shaozhou Tuhua 韶州土话/韶州土話: spoken at the border regions of Guangdong, Hunan, and Guangxi. This is an area of great linguistic diversity, and has not yet been conclusively described or classified.

In addition, the Dungan language (东干语/東干語) is a dialect of Mandarin spoken in Kyrgyzstan. However, it is written in the Cyrillic alphabet as a result of Soviet rule.

Quantitative similarity

A 2007 study compared 15 major urban dialects on two objective and two subjective criteria:[4]

Lexical similarity

Phonological regularity (regularity of sound correspondences, not direct phonological similarity)

Subjective intelligibility

Subjective similarity

Major north+central vs. south split

Generally the top-level split put Northern, New Xiang, and Gan in one group and Min (samples at Fuzhou, Xiamen, Chaozhou), Hakka, and Yue in the other group, except for phonological regularity, where the one Gan dialect (Nanchang) was in the Southern group and very close to Hakka, and the deepest phonological difference was between Wenzhounese (the southernmost Wu dialect) and all other dialects.

Lack of clear splits within the north+central area

Changsha (New Xiang) was always within the Mandarin group. No Old Xiang dialect was in the sample.

Taiyuan (Jin or Shanxi) and Hankou (Wuhan, Hubei) were subjectively perceived as relatively different from other Northern dialects, but were very close in subjective intelligibility. Objectively, Taiyuan had substantial phonological divergence but little lexical divergence.

Chengdu (Sichuan) was somewhat divergent lexically, but very little on the other measures.

Intermediate position of Wu, and unintelligibility of Wenzhounese

The two Wu dialects were closer to the Northern/New Xiang/Gan group in lexical similarity and strongly closer in subjective intelligibility, but closer to Min/Hakka/Yue in phonological regularity and subjective similarity, except that Wenzhou was farthest from all other dialects in phonological regularity. The two Wu dialects were close to each other in lexical similarity and subjective similarity, but not in subjective intelligibility, where Suzhou was actually closer to Northern/Xiang/Gan than to Wenzhou.

High divergence within Min

Fuzhou (Eastern Min) grouped only weakly with the Southern Min dialects Xiamen and Chaozhou on the two objective criteria, and was actually slightly closer to Hakka and Yue on the subjective criteria.

Closeness of the southernmost dialect areas

Hakka and Yue grouped closely together on the three lexical and subjective measures, but not in phonological regularity.

Local classifications

Generally, when referring to a local dialect in everyday speech, the speaker will refer to the dominant city in the region as a marker of the dialect as a whole. For example, a Wu speaker would not ask a fellow Wu speaker if they speak "Wu", but would rather ask whether or not they speak the dialect from Suzhou or Hangzhou, known as Suzhouhua and Hangzhouhua, respectively, in Chinese. Generally dialects are branded according to cities, geographical regions, or provinces. This method of informal classification is commonly used in spoken language. Provinces whose dialects are more homogeneous within its boundaries, such as Shaanxi, Shanxi, Shandong, Hebei, Hunan, Jiangxi, Sichuan, etc. tend to refer to their own dialects by the name of the province (although sub-dialects exist and can be referred to locally by the name of a city). In more diverse provinces such as Fujian, dialects are informally classified by mutual intelligibility into Min Nan (闽南话), Min Dong (闽东话), and Min Bei (闽北话); in Zhejiang, where there is vast variance in spoken language, dialects are generally classified by cities or counties – as such, no singular "Zhejiang dialect" exists. An area with widespread homogeneity in spoken language is the three provinces of Northeastern China, whose spoken language is collectively known as Northeastern Mandarin, or Dongbei Hua (东北话) in Chinese.