Graphical Analysis of Ranked Counts (of Words)
Anita Parunak · Journal of the American Statistical Association · 1979
Items that are counted may lack a natural ordering or classification by factors. This article develops a data-analytic technique for dealing with the distribution, often very long-tailed, of such counts. The technique, applied to counts of words from large texts in Greek, French, and English, discovers a set of word frequency distributions dependent on word length. A form of the incomplete gamma function estimates the expected number of words of a specified length.