Mandarin, Cantonese, Shanghainese, and Hokkien are commonly called dialects of Chinese. Yet the differences between them can be greater than those separating several officially recognized European languages. Understanding why reveals something fascinating about how we define languages.
Most people think of Chinese as one language spoken in different regional dialects. Mandarin is the official standard in mainland China, while Cantonese, Shanghainese, Hokkien, Hakka, and other varieties are commonly described as regional forms of the same language.
But there is an interesting complication: many Chinese "dialects" are not mutually intelligible.
A Mandarin speaker generally cannot understand a conversation conducted entirely in Cantonese, Shanghainese, or Hokkien without previous exposure or study. These varieties differ substantially in pronunciation and vocabulary, as well as in some aspects of grammar.
Now consider the languages of the former Yugoslavia. Serbian, Croatian, Bosnian, and Montenegrin are recognized as distinct national standard languages. Yet their speakers can generally understand one another with little difficulty.
This creates a fascinating contrast. Some officially recognized languages are extremely similar, while some varieties traditionally classified as dialects are dramatically different.
So what exactly makes something a language rather than a dialect?
The answer involves more than vocabulary and grammar. History, cultural identity, education, and language standardization also play important roles. Recognizing linguistic differences does not imply that communities must have separate national or political identities. Language classification and national identity are related but fundamentally different questions.
Chinese is a diverse family of related languages
Chinese belongs to the Sino-Tibetan language family. Its varieties form the Sinitic branch, which linguists commonly divide into several major groups.
Depending on the classification system, Chinese has approximately seven to ten major groups, with numerous regional varieties within each.
| Group | Well-known varieties | Main regions |
|---|---|---|
| Mandarin | Standard Mandarin, Sichuanese | Northern and southwestern China |
| Wu | Shanghainese, Suzhounese | Shanghai, Zhejiang |
| Yue | Cantonese, Taishanese | Guangdong, Hong Kong |
| Min | Hokkien, Teochew, Fuzhounese | Fujian, Taiwan |
| Hakka | Hakka varieties | Guangdong, Fujian, Jiangxi |
| Xiang | Hunanese | Hunan |
| Gan | Gan varieties | Jiangxi |
| Jin | Jin varieties | Shanxi |
| Huizhou | Hui varieties | Southern Anhui |
| Pinghua | Pinghua varieties | Guangxi |
The exact classification is not universally agreed upon. For example, Jin is sometimes grouped with Mandarin, while Min contains several highly divergent varieties.
Even belonging to the same group does not guarantee mutual intelligibility. Hokkien and Fuzhounese both belong to Min Chinese, yet their speakers generally cannot understand one another without previous exposure.
Some neighboring communities in southern China also speak varieties that can be difficult for outsiders, including speakers of other Chinese varieties, to understand.
Encyclopaedia Britannica describes Chinese as a collection of varieties popularly called dialects but usually classified as separate languages by scholars. It compares their differences to those among the modern Romance languages, such as Spanish, Italian, and French.
Describing Chinese as a group of related languages is therefore not an unusual linguistic argument. It is an established way of understanding its diversity.
What makes a language different from a dialect?
One of the most useful concepts in linguistics is mutual intelligibility, or the ability of speakers of different varieties to understand one another without prior study.
Consider American and British English. They differ in pronunciation, vocabulary, and some grammatical conventions, but speakers can generally communicate without difficulty. Most people recognize them as varieties of English.
Now consider Spanish and Portuguese. These languages share a common ancestor and many similarities, but understanding between their speakers is less reliable, especially in spontaneous speech.
Mandarin and Cantonese present a comparable situation. Although both belong to the Chinese linguistic tradition, knowing one does not automatically allow someone to understand the other.
The same applies to Mandarin and Shanghainese, or Mandarin and Hokkien.
These are not merely different accents. They are distinct linguistic systems that have developed over centuries, with substantial differences in their sound systems, vocabulary, and certain grammatical structures.
Mutual intelligibility is not a perfect test, however. A Mandarin speaker who regularly watches Cantonese television may understand considerably more Cantonese than someone who has never encountered it. Exposure, education, and regional contact can significantly influence comprehension.
Languages can also form dialect continua, where neighboring varieties are relatively easy to understand, but more distant varieties become increasingly different.
Nevertheless, mutual intelligibility provides a useful starting point for understanding why many Chinese varieties can reasonably be considered separate languages.
Mandarin and Cantonese differ in more than pronunciation
To appreciate the differences between Chinese varieties, consider some everyday expressions.
| English | Mandarin | Cantonese |
|---|---|---|
| I | 我 (wǒ) | 我 (ngo5) |
| He or she | 他 / 她 (tā) | 佢 (keoi5) |
| Eat | 吃 (chī) | 食 (sik6) |
| Not | 不 (bù) | 唔 (m4) |
| To be | 是 (shì) | 係 (hai6) |
| What | 什么 (shénme) | 乜嘢 (mat1 je5) |
| Now | 现在 (xiànzài) | 而家 (ji4 gaa1) |
Some words use the same characters but have different pronunciations. Others use entirely different words and characters.
The differences become clearer when comparing complete sentences.
English: What do you want to eat?
Mandarin: 你想吃什么？
Nǐ xiǎng chī shénme?
Cantonese: 你想食乜嘢？
Nei5 soeng2 sik6 mat1 je5?
Both sentences express the same meaning and share elements of their grammatical structure. However, their vocabulary and pronunciation differ substantially.
Cantonese also uses grammatical particles, aspect markers, and other constructions that do not correspond directly to everyday Mandarin usage.
The two varieties share a historical foundation, but that does not make them interchangeable.
The surprising comparison with the former Yugoslavia
Perhaps the most striking illustration of how language classification works comes from southeastern Europe.
During much of the twentieth century, Serbian, Croatian, Bosnian, and Montenegrin speech communities were associated with a shared standard language commonly known as Serbo-Croatian.
Following the dissolution of Yugoslavia, four distinct national standard languages became established: Serbian, Croatian, Bosnian, and Montenegrin.
Each has its own official status, standardized conventions, literary traditions, and relationship with national identity. Differences exist in vocabulary, pronunciation, spelling, and preferred expressions. Serbian also uses both Cyrillic and Latin scripts, while Croatian uses Latin.
Yet speakers of these four standard varieties can generally communicate without translation.
A Croatian speaker can usually watch Serbian television and understand what is being said. A Bosnian speaker can generally converse with a Montenegrin speaker without needing to study Montenegrin as a foreign language.
The differences are real, but mutual intelligibility remains very high.
This situation contrasts with Mandarin and Cantonese, whose speakers often need to use a shared language, such as Standard Mandarin, to communicate.
| Language varieties | Typical spoken intelligibility without prior study |
|---|---|
| Serbian and Croatian | Very high |
| Croatian and Bosnian | Very high |
| Serbian and Montenegrin | Very high |
| Mandarin and Cantonese | Low |
| Mandarin and Shanghainese | Low |
| Mandarin and Hokkien | Low |
| Hokkien and Fuzhounese | Low |
These are broad qualitative descriptions rather than precise numerical measurements. Actual comprehension depends on education, exposure, and the specific varieties involved.
Nevertheless, the central contrast is clear.
Serbian, Croatian, Bosnian, and Montenegrin are recognized as separate national standard languages despite their close linguistic relationship. Mandarin, Cantonese, and Hokkien are often called dialects despite being far less mutually intelligible.
This does not mean that the national languages of the former Yugoslavia should be considered invalid or that their distinctions are unimportant. Their separate standards and identities are well established.
Rather, the comparison demonstrates that linguistic similarity is not the only factor determining whether a variety is recognized as a language.
This is not unique to Chinese
Similar situations exist throughout the world.
Norwegian, Swedish, and Danish are recognized as separate languages, but their speakers can often understand one another to varying degrees. Norwegian and Swedish are particularly close in many respects, although comprehension varies by dialect and individual exposure.
Hindi and Urdu provide another interesting example. Their everyday colloquial forms can be highly mutually intelligible, while their formal standards differ in vocabulary, writing systems, and literary traditions.
Arabic presents a situation more comparable to Chinese. Modern Standard Arabic provides a common formal written language across much of the Arab world, while everyday spoken varieties can differ dramatically. Moroccan Arabic and Iraqi Arabic, for example, can be difficult for their respective speakers to understand without prior exposure.
German offers another example. Some Swiss German varieties differ considerably from Standard German, yet they are generally grouped within the broader German linguistic tradition.
These examples demonstrate that there is no universal rule under which closely related varieties must always be classified as dialects, or highly divergent varieties must always be recognized as separate languages.
Sometimes closely related varieties develop independent national standards. Sometimes significantly different varieties remain part of a shared linguistic and cultural identity.
Why are Chinese languages still called dialects?
There are important historical, cultural, and linguistic reasons for the terminology.
First, Chinese-speaking communities share a long literary and cultural history. Their languages developed from related historical forms of Chinese, and many communities have maintained connections through literature, education, and cultural exchange.
Second, Standard Mandarin functions as a common language across mainland China and is also central to education and public communication in Taiwan. Speakers of different regional varieties frequently learn Mandarin and use it to communicate beyond their local communities.
Third, the Chinese term 方言 (fāngyán), commonly translated as "dialect," does not correspond perfectly to the everyday English meaning of that word. It can refer broadly to regional forms of speech, including varieties that are mutually unintelligible.
Finally, language labels are shaped by historical conventions and shared cultural identities. A common identity can encompass considerable linguistic diversity, just as separate identities can develop around closely related languages.
The term "Chinese dialect" is therefore understandable within its historical and cultural context.
The difficulty arises when English-speaking readers interpret it as meaning a minor variation of the same spoken language.
Doesn't everyone use the same Chinese characters?
One reason Chinese varieties appear more unified than they sound is their shared writing tradition.
A Cantonese speaker and a Mandarin speaker may struggle to understand each other's speech, yet both can read Standard Written Chinese if they have learned it.
However, that does not mean the written standard directly represents every spoken variety in the same way.
Standard Written Chinese is largely based on Mandarin vocabulary and grammar. Speakers of other Chinese varieties typically learn this written standard through formal education, even when it differs from their everyday speech.
Cantonese provides a particularly clear example.
English: He is not at home.
Standard Written Chinese: 他不在家。
Colloquial written Cantonese: 佢唔喺屋企。
Both sentences express the same meaning, but their vocabulary and grammatical conventions differ.
A person familiar only with Standard Written Chinese may not recognize every expression used in colloquial written Cantonese.
Other Chinese varieties also have distinctive vocabulary and grammatical structures, although their written traditions and degree of standardization differ.
A shared writing system can facilitate communication across linguistic boundaries, but it does not eliminate those boundaries.
The reverse is also true. Serbian and Croatian can use different writing conventions while remaining highly mutually intelligible in speech.
Writing systems and spoken languages are related, but they are not the same thing.
Why these distinctions matter for translation and AI
The differences between Chinese varieties are not merely academic. They have practical consequences for translation software, speech recognition, subtitles, and artificial intelligence.
A speech recognition model trained primarily on Mandarin cannot automatically be expected to understand Cantonese or Hokkien accurately. These languages have different sound systems, vocabulary, and linguistic conventions.
Similarly, translating text into Standard Written Chinese does not necessarily produce the language a Cantonese speaker would naturally use in everyday conversation.
A translation service that advertises support for "Chinese" might mean Standard Mandarin, Standard Written Chinese, Cantonese, or some combination of these. The distinction matters because support for one does not guarantee support for the others.
For example, an AI translation system may need language-specific training and evaluation to produce natural Cantonese rather than Mandarin-based written Chinese. Speech recognition systems must also account for different pronunciations, vocabulary, and sometimes entirely different linguistic structures.
The challenge extends beyond the best-known varieties.
Hokkien, Hakka, Teochew, Fuzhounese, and other regional languages have substantial speech communities but often receive less attention from mainstream language technology.
Supporting these languages requires more than treating them as variations of Mandarin. It can involve collecting appropriate training data, developing writing conventions, evaluating native-language output, and working with speakers who understand regional differences.
This is especially important for real-time translation, where two people speaking different languages need to communicate naturally without relying on a shared language.
Recognizing linguistic diversity can improve translation accuracy, accessibility, and language preservation. It also helps technology developers describe their language coverage more accurately.
So is Chinese one language or many?
The answer depends on how the word "Chinese" is being used.
From a historical and cultural perspective, Chinese can serve as an umbrella term covering numerous related linguistic varieties that share important literary and cultural traditions.
From the perspective of comparative linguistics, it is often more informative to describe Chinese as a group of related Sinitic languages within the Sino-Tibetan family.
Both descriptions can be useful because they answer different questions.
Calling Cantonese, Hokkien, or Shanghainese a separate language in the linguistic sense does not deny its connection to Chinese history or culture. Nor does it imply anything about the nationality or political identity of its speakers.
Likewise, recognizing Serbian, Croatian, Bosnian, and Montenegrin as distinct national standard languages does not require assuming that their linguistic differences are greater than they are.
Language classification involves at least two dimensions: how different varieties are linguistically, and how communities identify, standardize, and use them.
These dimensions do not always align.
Conclusion: A language is more than a linguistic classification
The distinction between a language and a dialect may seem straightforward, but Chinese and the languages of the former Yugoslavia demonstrate how complicated it can become.
Serbian, Croatian, Bosnian, and Montenegrin are distinct national standard languages despite their high mutual intelligibility. Mandarin, Cantonese, Hokkien, and Shanghainese are often described as dialects despite major linguistic differences that can prevent easy communication between their speakers.
Neither situation can be explained by mutual intelligibility alone.
History, education, literary traditions, cultural identity, and standardization all influence how languages are classified and named.
The important takeaway is not that some officially recognized languages should lose their status, or that every regional Chinese variety must receive a particular label.
It is that linguistic diversity does not always correspond to the language categories we use in everyday life.
Chinese is a particularly compelling example. Behind a single familiar name lies an extraordinary range of related languages, many of which differ from one another as substantially as languages recognized separately elsewhere in the world.
Understanding that diversity gives us a more accurate picture of Chinese and a greater appreciation of the complexity of human language.
References and further reading
- Encyclopaedia Britannica: Chinese Languages. Overview of Chinese linguistic diversity, mutual intelligibility, and major Sinitic language groups.
- Jin Liu: A Historical Review of the Discourse of Fangyan in Modern China. Research on Chinese regional languages, terminology, and the historical development of language identity.
- Michele Mannoni: Chinese Dialects/Dialects of Chinese. Academic discussion of terminology and the distinction between dialects and languages in Chinese linguistics.
- Pluricentricity in the Classroom: The Serbo-Croatian Language. Research exploring the relationship, intelligibility, and standardization of Bosnian, Croatian, Montenegrin, and Serbian.
- Ljubica Djordjević: The Legal Status of Languages That Emerged from Serbo-Croatian. Analysis of the official language standards that developed following Yugoslavia's dissolution.