Can Mandarin speakers understand Shanghainese?
The short answer is no. The interesting answer is that comprehension runs almost entirely in one direction, and that the reason is not linguistic — it is that one group had to learn the other language at school and the other group did not.
Tap 🔊 to hear an example. Shanghainese examples play the recordings from Verbavia’s course, made with a Shanghainese voice; Mandarin examples play a Mandarin voice.
The short answer
Shanghainese and Mandarin are not mutually intelligible. Wikipedia states it flatly: Shanghainese, “like the rest of the Wu language group, is mutually unintelligible with other varieties of Chinese, such as Mandarin Chinese”.
A Mandarin speaker with no exposure to Wu, listening to a Shanghainese conversation, will pick out the occasional word and lose the sentence. This is not a matter of accent or of speed. The commonest words are different words.
The experiment
This is usually argued by assertion. It has also been measured. Chaoju Tang and Vincent van Heuven at Leiden tested Chinese varieties functionally — playing speakers of one variety to listeners of another and scoring how much they could actually translate — and published the results in Lingua in 2009.
Their sample did not include Shanghai itself. The northern Wu variety they tested was Suzhou, which sits in the same branch and was Wu’s prestige variety before Shanghai took that role, so it is a close but not identical proxy. With that caveat, here is the sentence-level result, scored as the percentage of target words the listeners correctly translated:
| Who is speaking | Who is listening | Understood |
|---|---|---|
| Suzhou (northern Wu) | Beijing (Mandarin) | 18% |
| Beijing (Mandarin) | Suzhou (northern Wu) | 87% |
| Wenzhou (southern Wu) | Beijing (Mandarin) | 2% |
| Beijing (Mandarin) | Wenzhou (southern Wu) | 62% |
Eighteen per cent against eighty-seven. That is not a language pair struggling slightly in one direction; it is a near-total failure one way and near-complete success the other. And the southern Wu row is more striking still: two per cent, which is indistinguishable from not sharing a language at all.
Why it is not mutual
Here is the part most discussions miss, and it matters. Mutual intelligibility is supposed to be a property of the languages. What that table measures is partly a property of the people.
The authors are explicit about it. Most of their listeners already spoke some Standard Mandarin — around sixty per cent said they could speak it — and they note this “may have implications for the interpretation of the results”. Of course it does. The Wu listeners understood Beijing speech at eighty-seven per cent because they had been taught Mandarin since primary school. The Beijing listeners had never been taught anything about Wu.
So the asymmetry is not evidence that Wu is somehow closer to Mandarin than Mandarin is to Wu, which would be incoherent. It is evidence of an education system. Strip out the schooling and you would be left with two languages that are hard work in both directions — which is what the eighteen per cent figure is closer to showing.
This is worth holding onto whenever someone says Shanghai people understand Mandarin fine, so the two must be close. They understand it fine because they learned it.
What a Mandarin speaker does catch
Not nothing. The overlap is real and it is concentrated in particular places:
- Anything written. The characters are shared, so a Mandarin reader reads Shanghainese text with reasonable success and hears none of it. This is the single biggest source of the confusion about how close the two are.
- Formal and technical vocabulary. The bookish layer of Shanghainese was borrowed from prestige Mandarin over centuries, so abstract and institutional words often match — which is exactly what the literary and colloquial readings page is about.
- Names, numbers in context, loanwords. Enough to guess a topic, rarely enough to follow an argument.
What they do not catch
The everyday layer, which is most of a conversation:
- The pronouns. Not one of the six is the Mandarin word, and they are in every sentence — see the pronouns page.
- The commonest verbs and function words. 勿 veh ‘not’ rather than 不, 晓得 shiau teh ‘to know’ rather than 知道, 啥 sa ‘what’ rather than 什么.
- The sound system. Wu kept the voiced initials of medieval Chinese and the checked tone; Mandarin lost both. A Mandarin ear has no category for the voiced series, which is set out on the pronunciation page.
- The tones. Not just different values but a different system, in which the first syllable of a word decides the melody of the whole of it.
How far apart is that, really?
The standard comparison is to the Romance languages, and Wikipedia’s article on the varieties of Chinese puts it about as strongly as anyone does: 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”.
Treat that as an analogy rather than a measurement — nobody has a ruler that converts between language families. But the shape of it is right, and it is a more useful mental model than dialect. A Mandarin speaker meeting Shanghainese is in roughly the position of a Spanish speaker meeting French: obviously related, recognisable in writing, and not something you can follow by ear without study.
Then why does everyone call it a dialect?
Because the Chinese word 方言 is translated that way, and the English word carries an implication the Chinese one does not — that the thing is a regional flavour of the standard and comprehensible to its speakers. That mistranslation does a lot of damage to this exact question, and it has a page of its own: what a topolect is, along with the argument about whether Shanghainese is a language, on the language or dialect page.
The practical upshot
If you speak Mandarin and you are going to Shanghai, you will be fine: Mandarin is the working language of the city and everyone under sixty speaks it. What you will not do is understand the conversation at the next table, and no amount of Mandarin will change that.
If you want to understand that conversation, you are learning a second language rather than adjusting to an accent. The good news is that a great deal of it transfers — the writing, a lot of the vocabulary, the grammar's basic shape — and what does and does not is set out on the page for Mandarin speakers.
One more number, for perspective on who you would be talking to. About forty per cent of Shanghai residents under eighteen cannot speak the language fluently. The conversation at the next table is increasingly likely to be in Mandarin anyway — which is the subject of the page on whether Shanghainese is dying.
Sources
- Wikipedia: Shanghainese — “mutually unintelligible with other varieties of Chinese, such as Mandarin Chinese”; about 40% of residents under 18 cannot speak it fluently
- Tang, Chaoju and van Heuven, Vincent J. (2009). Mutual intelligibility of Chinese dialects experimentally tested. Lingua 119, 709–732 — the functional intelligibility tables, and the note that most listeners already spoke Standard Mandarin
- Wikipedia: Varieties of Chinese — “differing at least as much as different Romance languages and perhaps even as much as Indo-European languages as a whole”
- Wikipedia: Wu Chinese — Wu preserves the voiced initials of medieval Chinese and the checked tone