回复: 如何解读Readability Analyzer中的数据
Most readability statistics are rough estimation of text difficulty. They should not be recommended as sole criteria for the readability of texts. It is always good to know the weaknesses of the readability formulae.
A garden path sentence (e.g.The horse raced past the barn fell) or a literary text can easily confuse a computer AND human readers too. Readability tests are normally whole text based; exceptional and irregular sentences will be averaged or neutralized in the overall score.
Interestingly, however rough the statistics are, we've found, in an earlier empirical study, that readability statistics are quite robust in relation to text difficulty as compared with human judgment. Out of over a hundred linguistic variables, readability stats are always highly correlated with the difficulty scores given by experienced human raters.
Considering the weaknesses of readability scores, we incorporated multiple scores and values (e.g. TTR, STTR, lemma TTR etc) in Readability Analyzer to triangulate the text or lexical complexity of texts. Nonetheless, we still believe that those values are but approximation of text difficulty, which should be corroborated otherwise.