Cantonese-Mandarin Sound Correspondences

I heard Cantonese from a young age but never learned to speak it, and I learned Mandarin at school. So when I attempt to say a Cantonese word I don't already know, I'm often trying to take a Mandarin character and map it to its most likely Cantonese equivalent. And vice versa.

This tool shows how successful this approach is likely to be in either direction, with tables demonstrating the phonological correspondences between the two languages, followed by a statistical analysis. Data is from the Unihan database which contains Mandarin (pinyin) and Cantonese (jyutping) readings.

Prepared by Michelle Fullwood.

Select a direction and component to explore how initials, finals, tones, and full syllables map between the two languages.

How predictable are the correspondences overall?

If you always guess the most likely correspondence, how often would you be right?

Across 27,522 character readings:

Strategy M→C C→M
Guess most likely initial 73.7% 64.0%
Guess most likely final 47.8% 65.5%
Guess syllable from most likely initial + most likely final 36.0% 43.2%
Guess syllable from most likely whole syllable 57.4% 68.5%
Guess most likely tone 64.5% 80.6%

This strategy is more successful in the Cantonese → Mandarin direction overall. Generally, Cantonese has been more conservative in preserving finals and tones from Middle Chinese, hence there are more possibilities for Cantonese finals and tones than in Mandarin. Choosing from among more options is always going to be more difficult.

The exception is in the initials, where Mandarin actually has more distinctions than Cantonese. For example, Mandarin palatalized its velars (g/k/h) and sibilants (z/c/s) into j/q/x before front vowels, creating a third series that Cantonese lacks. So going from Cantonese → Mandarin, each Cantonese initial like /g/ or /z/ could either stay as-is or become /j/ — more options to choose from, and thus harder to predict.

Since the palatalization depends on the following vowel, knowing the final helps predict the initial. Similarly, knowing the initial helps predict the final. Which is why guessing the syllable as a unit does much better than guessing the initial and final independently.

Having said that, this guessing approach isn't a dead-end. There are reliable correspondences, you just need to have an instinct (or memory) for which. Studying the tables may help, but ultimately there is no substitute for actual language practice — listening or speaking.