Skip to content

Reading Shanghainese in Characters When You Only Know Mandarin Readings

A character on the page is a written unit, not a pronunciation. What happens when you can read a Shanghainese sentence in Mandarin but cannot say it in Wu.

By Published 5 min read

Reign Creative LLC makes Learn Shanghainese Wu Chinese, which this article links to, so we have a commercial interest in it. How we write and check articles.

Key takeaways

  • A Chinese character encodes a written unit, not a single sound, and the same character is read differently by speakers of different Sinitic varieties and by Japanese, Korean and Vietnamese readers of the same script.
  • The Unicode Han Database records readings in separate per-variety fields - kMandarin, kCantonese, kJapaneseOn, kJapaneseKun, kKorean, kVietnamese and others - which makes the variety-specific nature of a reading explicit in the standard itself.
  • There is no Wu or Shanghainese reading field among those Unihan properties, so the mapping a Shanghainese learner needs is not available from the largest standard character reference.
  • Because Shanghainese is classified within Wu rather than under Mandarin, guessing a Wu pronunciation from a Mandarin reading produces systematic errors rather than an accent.
  • The practical fix is to learn characters paired with a Shanghainese romanisation from the start, so that reading and pronouncing are acquired together instead of one being inferred from the other.

Here is a specific, common and rarely addressed problem. You open a page of Shanghainese written in Chinese characters. You understand it. You cannot say a single line of it out loud in Shanghainese, because the only character-to-sound mapping you own is a Mandarin one.

Nothing is wrong with your reading. The written form simply never carried the information you are looking for.

A character is a written unit, not a sound

The clearest way to see this is to look at how the character set is documented rather than at how it is taught.

The Unicode Han Database — the reference that assigns properties to the characters in the Unicode standard — does not store "the pronunciation" of a character. It stores readings in separate, explicitly labelled fields for different reading traditions: kMandarin for Mandarin, kCantonese for Cantonese in jyutping, kHanyuPinyin for readings drawn from a particular dictionary, kJapaneseOn and kJapaneseKun for the two Japanese reading types, kKorean, kVietnamese, and kTang for historical Tang-era readings.

That structure is the answer to the question in miniature. A character is a shared written unit that many reading traditions attach different sounds to. The standard does not treat a Mandarin reading as the pronunciation with other varieties as deviations; it treats each as a separate property of the same written form.

The documentation is candid about the limits, too. It notes that in most cases a character has only one reading, or one important reading, but that there are numerous exceptions — and that pronunciation is tied to the word a character forms part of, varying over time and from place to place, even within a single language.

The gap that affects Shanghainese specifically

Now the uncomfortable part. Among those Unihan reading fields there is one for Mandarin and one for Cantonese, and there is nothing for Wu or for Shanghainese.

That is not a criticism of the standard, which records the reading traditions that have been systematically compiled. It is a precise statement of the learner's problem: the mapping you need — this character, in Shanghainese, is pronounced this way — is not something you can look up in the same place you would look up a Mandarin or Cantonese reading. It has to come from material specifically prepared for Wu.

And Wu is where Shanghainese sits. Glottolog places Shanghainese in a chain that runs through Wu Chinese, Taihu and Suzhou, and treats Wu Chinese as a Sinitic language in its own right with its own ISO code. Our overview of Wu Chinese and where Shanghainese sits within it covers the family in more detail. The point for reading is structural: Shanghainese is not downstream of Mandarin, so Mandarin readings are not a slightly-off version of Shanghainese readings. They are a different branch's readings.

What goes wrong when you guess

Inferring a Wu pronunciation from a Mandarin one is not like guessing an accent. It produces errors with a pattern, and the pattern is worth naming because it explains why the approach fails so completely rather than approximately.

  • Sound inventories differ. Wu varieties retain distinctions Mandarin lost and lack some Mandarin has. There is no consistent substitution rule to apply.
  • Tone systems differ in kind. Shanghainese tone behaviour is dominated by what happens when syllables combine, not by the tone of each syllable in isolation, which means a word-by-word reading can be wrong even when each syllable would be right alone.
  • Vocabulary is not identical. Some everyday Shanghainese words are not the Shanghainese reading of the Mandarin word at all; they are different words, sometimes written with different characters, sometimes with no settled character.
  • Grammar diverges. A sentence understood through Mandarin word order may be read correctly and still not match how the sentence is actually said.

Shanghainese vs Mandarin differences goes through this in the direction most learners care about — what your Mandarin buys you and what it does not — and the honest summary is that it buys meaning and buys very little sound.

The three-way distinction worth holding

There is a further complication that causes a lot of confusion, and it is easier once named. Three different things get called "written Chinese" in a conversation about Shanghainese:

Standard Written Chinese is the written standard used across Chinese-speaking regions, broadly aligned with Mandarin grammar and vocabulary. Most published text is this.

Characters used to write Shanghainese words is a different exercise: representing what a Shanghainese speaker actually says, including words that Standard Written Chinese would express differently. Practice here is less uniform, and some morphemes have no universally agreed character.

Romanisation is a transcription system that records the Shanghainese pronunciation directly, and it is the only one of the three that reliably tells you how to say something.

Because no single romanisation for Shanghainese has the standardised status that pinyin has for Mandarin or jyutping has for Cantonese, the choice of system is a real decision — romanising Shanghainese works through the options and the trade-offs. The practical requirement is consistency rather than correctness: one system, used throughout your material, so that one character maps to one transcription in your memory.

Learning both at once

The fix follows directly from the diagnosis. If pronunciation cannot be recovered from the character, then the character must be learned with the pronunciation attached from the first exposure — not read first and corrected later.

Learn Shanghainese Wu Chinese, which is our own app, is built around that pairing: Chinese characters and Shanghainese romanisation are presented together specifically so you can read and pronounce without guessing from Mandarin, and the course includes dedicated lessons on tone sandhi and connected speech rather than leaving sandhi as an advanced topic. The app's own FAQ makes the entry requirement explicit — you do not need to know characters beforehand, because they are introduced alongside romanisation and English meanings.

Three habits make that pairing stick:

  1. Never store a character without its Shanghainese transcription. A character alone in your notes will be re-read in Mandarin the next time you see it.
  2. Practise in phrases, not single characters. Sandhi makes the phrase the real unit of pronunciation.
  3. Keep a Mandarin column if you want one, but keep it visibly separate. Comparison is useful; blending the two mappings is exactly the failure this article is about.

The rest of the language learning articles cover the general study methods — spaced review, listening practice, plateau-breaking — that apply once the reading problem is solved. This one is worth solving first, because everything downstream of it inherits the error.

Good to know

Frequently asked questions

Is Shanghainese written in Chinese characters?
Characters are the script used across Sinitic varieties, so Shanghainese material is normally presented in them. What characters do not carry is a Shanghainese pronunciation: the same written form is read differently by speakers of different varieties, and Shanghainese readings are not recoverable from the written form on its own.
Does a Chinese character have one pronunciation?
Not across varieties, and often not within one. The Unicode Han Database keeps readings in separate fields for Mandarin, Cantonese, Japanese, Korean, Vietnamese and other traditions, and its documentation notes that while most characters have one important reading there are numerous exceptions, and that pronunciation varies over time and from place to place even within a language.
Why can I read a Shanghainese sentence but not say it?
Because you are reading it in a variety you already know. If your character-to-sound mapping was built through Mandarin, you will recover the meaning of a Shanghainese sentence written in characters and produce Mandarin sounds while doing so. The characters transmit the words; the pronunciation has to be learned separately for each variety.
So how do learners get Shanghainese pronunciations?
By pairing characters with a romanisation of the Shanghainese reading and learning both together. Because no single romanisation is universally standardised for Shanghainese, the practical requirement is that whatever course you use stays internally consistent, so that one written form maps to one transcription throughout your study.

Sources

  1. Unicode Standard Annex #38: Unicode Han Database (Unihan) (opens in a new tab)Unicode Consortium — accessed
  2. Wu Chinese (opens in a new tab)Glottolog — accessed
  3. Shanghainese (opens in a new tab)Glottolog — accessed

About the author

— The independent app studio that builds the apps covered on this site and publishes every article here. On Google Play it publishes as Reign Collective Apps.

We use AI tools to help research, draft and edit our articles. Each article lists the sources its factual claims rely on, and details about our apps come from their live Google Play listings. No article on this site has been reviewed by an independent expert unless the reviewer is named on it.

Spotted an error? Tell us — how corrections are handled is set out in our editorial policy.

Keep reading