For English (and some other languages), it's quite easy to find online tools that give a readability index of a specific text. Naturally, this doesn't necessarily match actual experience of reading the text, but it does seem to be reasonably accurate. Here's an example of such a tool.

My question is: Are there any similar tools for Chinese?

If not, are there any other practical ways of automatically estimating the difficulty of a Chinese text? I have tried crude methods such as using an annotator that can show vocabulary at different HSK levels, which gives a clue to the difficult of the words in the text at least, but that's not really what I'm after. However, if anyone knows of partial solutions to the question, that would be interesting anyway.

  • Whatever you method you invent, it will certainly be better than the Flesch-Kinkaid method for analyzing English (which literally just counts words/sentence and syllables/word). Oct 1 '13 at 17:42
  • Perhaps I should have chosen a better example. There are at least quite a lot of text analysis tools available for English and I mostly chose one that looked like what I had used before.
    – Olle Linge
    Oct 2 '13 at 0:28

I'm not sure if this is what you are after:


  • Not really, because it only gives number of characters and unique characters per sentence. This is a small indication of difficulty, but not if analysed per sentence as is the case here. However, the site you linked to also links to another site which looks more promising, posting separate answer about it!
    – Olle Linge
    Oct 2 '13 at 0:30
  • 1
    Please try not to make link only answers, it's better to give at least a short description of what the link is about. People need to go to the link to find the real answer, plus the link might stop working in the future. Oct 2 '13 at 13:12
  • 见过扯淡的,但没见过这么扯淡的。这个问题的答案是一个算法和相关的应用,难道我要在这里把算法和论文贴出来? Oct 2 '13 at 13:46
  • NewLong, he's asking you to summarize the algorithm, not cut&paste the whole article. As congusbongus linked, it's part of Stackoverflow/Stackexchange's philosophy to prefer summaries+link over just-a-link. Oct 2 '13 at 17:03
  • @NewLong sorry if I came across as rude, perhaps if I just posted a link, I'd be less 扯淡? :) Oct 3 '13 at 1:20

I found one tool via the website that NewLong linked to in his/her answer: Chinese Vocabulary Profiler. It does the following:

  • A distribution of characters across ranges of frequency (how many characters fall within the 250 most common, how many within 251-500). Since difficult characters is at least part of the difficulty of a text, this is part of the answer.
  • A similar distribution, but now across the different types of vocabulary in the HSK (甲乙丙丁 and outside HSK). This is actually better, since it gives me as a teacher a better feel for how much might be beyond the students.
  • The same data for bigrams, trigrams and N-grams in the text. Now this is much closer to what I'm after, since words are the real unit of communication rather than individual characters. For instance, the tool displays words in the text sorted according to HSK, frequency and a number of other factors.

Pretty good as far as vocabulary goes. I'm guessing that grammar will be much harder (perhaps so hard that few people have even tried).

  • For this to be useful, it would be necessary to figure out what distributions at different levels look like. How many words beyond HSK X does an intermediate text have? What's considered difficult for students at level Y and so on. Still a good start, though.
    – Olle Linge
    Oct 2 '13 at 0:51
  • From what I've read, current research on Chinese readability is also using the same approach - the relative frequency of difficult words. As it's under active research, I doubt we'll see online tools soon. It's interesting that this approach is very different from Flesch-Kinkaid, which just counts sentence/syllable length. Oct 3 '13 at 1:14
  • Yes, perhaps the above tool is the best there is at the moment (and it's actually quite good). The difference between the system ought to be because in Chinese, word length is (almost) unrelated to difficulty (most words are disyllabic anyway), which is not the case in English. Still, I haven't done any research into this and I don't know how well Flesch-Kinkaid actually works (see comment to the original question).
    – Olle Linge
    Oct 3 '13 at 12:44

I've thinking for a while on a simple readability measure for Chinese and I have written a first approximation here: https://rolandcoeurjoly.wordpress.com/

Hope it helps.

  • "if you understand [99%] of the text, you can learn new words by context." I disagree a bit. I estimated that I only understood half of an Italian novel, but that was enough to enjoy the story and figure out a lot of new words.
    – 伟思礼
    Dec 9 '17 at 17:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.