[split] Why the VMS text is meaningful or meaningless?

[DonaldFisk]

I've added another page to my blog, on You are not allowed to view links. Register or Login to view..   What I've shown is that although the mean values of word pair frequencies are close to that expected by chance (i.e. there is no discernible word order), their variances are significantly higher than those in my generated manuscript.   In other words, the real manuscript's word frequencies are significantly less predictable.   To be precise, word frequencies follow a Poisson distribution for my generated manuscript, but a Gamma distribution for the real text.   I have speculated on a possible reason at the end of my article, but I'd like the opinion of an expert statistician.