There's plenty of good resources on Chinese character frequency available.

But I'm wondering about syllable frequency independent of characters, and also tone frequency both separate and in combination with syllable frequency.

It's not possible to properly derive this data from the character frequency due to characters having multiple readings.

  • 2
    Can the specific pronunciation of a character be derived semi-accurately from the surrounding chars., probably applying a Markov-chains model? If yes, probably having a program analyze a sufficiently large corpus of texts could yield such a count.
    – Joe Pineda
    Feb 5, 2014 at 6:08
  • 1
    The other option I'm thinking of is analyzing texts in pinyin or Wade-Giles that do indicate tone - that way the problem of which specific pronunciation to use has already been solved - except maybe in the case some texts marking specifically tone sandhi and others not (might be solvable with heuristics?). And even with texts that omit tone you could at least derive syllable frequency...
    – Joe Pineda
    Feb 5, 2014 at 6:11
  • Yes I'm assuming anyone that has compiled such frequency list would have two options. 1) Using an automated AI-based method such as Markov and 2) manually deciding the tones one by one on a large text corpus. 2 Would be preferred but since it's a lot more work it may not exist or might not be public. So I would also accept 1 but I'm not competent to create it myself. Feb 5, 2014 at 6:14
  • Yes just after submitting my comment I also thought about the effects of not just tone sandhi, but also unstressed syllables, two processes by which "dictionary tones" change into other "surface tones". Resources would be best if they deal with more of these effects, good if they state what they do and don't deal with, and poorest if they don't detail these properties. But all would still be useful to various degrees. Feb 5, 2014 at 6:16
  • 1
    I think having "surface tone" represented in orthography (especially the neutral-tone part) is much harder than it sounds, since I suspect it depends on many things, like talking speed, register, dialect etc. For example, it is my impression that tones tend to get neutralized less in Taiwan Mandarin than in Beijing Mandarin.
    – dainichi
    Feb 5, 2014 at 7:05


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.