Shakespeare Statistics


Authorship of Edward II 1590/91

Deutschland-Flaggeauf deutsch

Rolling delta is one function of the stylometric programme R Stylo and is immensely valuable in approaching authorship questions or collaborations. The delta values of the text in question are compared to those of suitable reference texts. The lowest value indicates the smallest difference in style. Introduced by Burrows in 2002 rolling delta examines a window of a certain size, determines the delta values and then moves to the next window according to a chosen step size. As it goes through the complete text step by step the lowest values create a curve that belongs to one or several reference texts and in this way gives the author of, or collaborators in the text. The results depend on the window size, the variables (words or character n-grams) and the number of variables. The culling value of 0 evaluates up to a thousand variables, N = 100 uses only those which are present in all files.
Each chart gives the name of the play, the window size, the step size, the type of the most frequent variables (here words = MF1W and character trigrams = MF3C). The culling value is also mentioned. It should be noted that MF3C accounts for more text than MF1W and, due to the larger population, is statistically more sound.

| 1 |
| 2 |
| 3 |
| 4 |
| 5 |
| 6 |
| 7 |
| 8 |
| 9 |
|10 |
|11 |
|12 |
|13 |
|14 |
|15 |
_top_
rolling delta

As you go through the various parameters you will notice that some information remains stable up to the point where populations become too small to yield reliable results. The smallest difference in style exists in the reference text of Shakespeare. Playwrights like Lodge and Munday could be excluded in preliminary investigations.