You might want to look into the "incidence of coincidence", which is the probability that letters from two randomly chosen text's are the same. It can be changed to measure a single message, and gives a distinctive difference for english vs. random texts.