What is a Chinese topolect?
English calls Cantonese and Shanghainese dialects of Chinese. The Chinese word behind that translation does not mean what the English word means, and almost every argument about whether these are languages is really an argument about the translation.
The word being translated
The Chinese term is 方言, fāngyán, literally place speech. It is old — it is the title of a Han-dynasty work surveying regional vocabulary — and it has been rendered into English as dialect for as long as anyone has been rendering it.
That translation imports something. In ordinary English, a dialect is a regional variety of a language that its other speakers can understand. Geordie is a dialect of English. If Shanghainese is a dialect of Chinese in that sense, then Mandarin speakers ought to understand it, and they do not — as the page on intelligibility sets out with numbers.
Mair’s objection, and his word
The sinologist Victor Mair set this out in 1991, in a paper called What Is a Chinese ‘Dialect/Topolect’?, and his statement of the problem is still the clearest one. English dialect, he wrote, normally means “one of two or more mutually intelligible varieties of a given language distinguished by vocabulary, idiom, and pronunciation” — and “the Chinese word simply does not mean that”.
He has a nice historical proof of it. Under the Qing, 方言 was applied by Chinese officials and scholars to Korean, Mongolian, Manchu, Vietnamese and Japanese. Whatever 方言 meant, it plainly never meant mutually intelligible with the standard, because nobody thought Manchu was.
His proposal was topolect — place plus speech, a literal calque, which he described as “an exact, neutral translation of fangyan” with the advantage of being neutral about how big the place is. He preferred it to John DeFrancis’s earlier regionalect partly on the aesthetic ground that topolect is entirely Greek in its parts while regionalect mixes Latin and Greek.
The point of the coinage is that it decides nothing. Calling Shanghainese a topolect says it is the speech of a place. It does not smuggle in a claim about whether a Beijinger can follow it, which is precisely the claim that dialect smuggles.
Did it catch on?
Partly, which is more than most coinages manage. The American Heritage Dictionary carries it, defining a topolect as “the language or speech of a particular place, such as a country, region, village, or valley”, with Sinitic languages such as Mandarin or Cantonese given as the special case. Mair has said he is satisfied with that definition, and reported that the Oxford English Dictionary’s editors intended to include it too.
In practice most linguists writing in English now sidestep the whole question with variety, or say Sinitic languages and let the classification carry the argument. Topolect survives as the precise option for when the distinction matters — which, on a site about Shanghainese, it constantly does.
What the official positions are
Two framings, both sourced, and they disagree:
- A family of languages. ISO 639-3 and Ethnologue assign separate language codes to the top-level groups — Mandarin, Wu, Yue, Min and the rest. On that view Shanghainese is a variety of Wu, and Wu is a language beside Mandarin rather than inside it.
- One language with regional variants. The official position of the government of the People’s Republic of China is that the varieties are variants of a single Chinese language.
This page is not going to adjudicate that, and it is worth saying why: the disagreement is not really about evidence. Both sides know that a Mandarin speaker cannot follow Shanghainese and that the two share a script and a literature. They weight those facts differently.
The Romance comparison, and the case against it
The standard analogy is that the Sinitic varieties differ about as much as the Romance languages do. Wikipedia puts it at least that strongly — local varieties are “often mutually unintelligible, differing at least as much as different Romance languages and perhaps even as much as Indo-European languages as a whole” — and notes the histories rhyme: an imperial language spread over local substrates about two thousand years ago, then diverging.
The counter-case is real and deserves stating properly rather than waving away. Latin’s descendants ended up as the national languages of separate states, each with its own standard, its own spelling and its own literature. The Sinitic varieties did not. They have shared a single script for two millennia — one that can be read aloud in any of them — along with a common literary canon and long periods of political unity. And their speakers overwhelmingly regard themselves as speaking one language. Wikipedia’s own article makes that case alongside the analogy: because speakers share a standard written form, the varieties “are popularly perceived among native speakers as variants of a single Chinese language”.
Speaker self-identification is not nothing. It is, in most of the world, one of the things that makes a language a language.
How big are the branches?
Wu is not a rounding error. By share of first-language speakers, the top-level Sinitic groups run roughly:
| Branch | Share of speakers |
|---|---|
| Mandarin | 65.7% |
| Min | 6.2% |
| Wu — the branch Shanghainese belongs to | 6.1% |
| Yue, including Cantonese | 5.6% |
| Jin | 5.2% |
| Gan | 3.9% |
| Hakka | 3.5% |
| Xiang | 3.0% |
Even the number of branches is a live question — Jin is sometimes folded into Mandarin, Pinghua into Yue — which is a reminder that classification here is a set of arguments rather than a settled inventory. Where Shanghainese sits inside Wu is on the Wu Chinese page.
So what should you call it?
For practical purposes: say what is true and skip the label. Shanghainese is the speech of Shanghai, it belongs to the Wu branch of Chinese, it shares a writing system with Mandarin, and a Mandarin speaker cannot understand it. Every one of those is checkable, and together they tell a reader everything the word dialect would have told them wrongly.
If you want one word, topolect is the accurate one and variety is the one people will recognise. What to avoid is dialect used as though it settled something, because in English it makes a claim about comprehension that happens to be false. The specific question of whether Shanghainese is a language is argued out on the language or dialect page.
Sources
- Mair, Victor H. (1991). What Is a Chinese “Dialect/Topolect”? Reflections on Some Key Sino-English Linguistic Terms. Sino-Platonic Papers 29
- Wikipedia: Varieties of Chinese — the criticism of “dialect” as a translation, the ISO 639-3 and PRC positions, the Romance analogy, the shared-script counter-case, and the branch percentages
- Language Log: Victor Mair on the American Heritage Dictionary’s definition of topolect, and the OED
- Wikipedia: Fangyan — the Han-dynasty survey of regional speech
- Wiktionary: topolect
- DeFrancis, John (1984). The Chinese Language: Fact and Fantasy — the earlier coinage “regionalect”