29-06-2020, 04:07 PM
If words are composed of arbitrary concatenations of several parts, then the distribution of each part does not have to be exactly binomial. As long as it is largely symmetric with a maximum in the middle, the process of combining them will make the result look close to binomial.
Note that (for example) a set of 18 prefixes, 36 stems and 18 suffixes generates 11,664 different words, slightly over the number of word types in the Voynich MS.
If their lengths are 0-3 , 1-5, 0-3 respectively, with individual distributions: (1,3,3,1) , (1,4,6,4,1) , (1,3,3,1), the resulting distribution would be 1-11 and binomial.
Unfortunately, the Voynich MS text does not really follow this model, but it show how a simple model can explain a couple of things.
Note that (for example) a set of 18 prefixes, 36 stems and 18 suffixes generates 11,664 different words, slightly over the number of word types in the Voynich MS.
If their lengths are 0-3 , 1-5, 0-3 respectively, with individual distributions: (1,3,3,1) , (1,4,6,4,1) , (1,3,3,1), the resulting distribution would be 1-11 and binomial.
Unfortunately, the Voynich MS text does not really follow this model, but it show how a simple model can explain a couple of things.