entropy splits in parts of lines?

[MichelleL11]

If the principle of filling out the nonsense parts was based on choosing rhyming words and phrases, as in my example above, then one would also expect the middles and ends of words to have much lower entropy (more predictability) than the beginnings of words, a statistical feature that we also observe to be present in the Voynich ms text. 

I don't know if the parts of the lines have been tested, but Rene Z. definitely tested the entropy of each individual character of words in the "stars" section (Quire 20) and unfortunately, he did not find what you are describing.  Instead, the beginnings of words were much more predictable than words from Latin and Italian, and the characters became increasingly more unpredictable as the word went along -- resulting in words that were collectively of the same "character" entropy per word as natural language.  But the distribution of the entropy values across each character differed from the natural languages tested (Latin and Italian).

Here is how Rene stated his conclusions.

• First-character and second-character entropy of Voynich MS text are significantly low, and thus especially also the initial-bigram entropy. However, in Voynichese, the third character of each word is as unpredictable as the second, and the fourth contains much more information than is the case in Latin or Italian
  

See the section called Entropy per Character on this page: You are not allowed to view links. Register or Login to view. for other data and discussion related to this.

Of course, this could turn out to be different if this were to be redone using single character substitutions for the bigrams that Koen identified as impacting overall second order entropy -- but as Koen's work has shown, it is hard to predict what the results on entropy is with alterations.  I would imagine this plays out at the character level entropy as well.  At least Rene's work does say that the initial bigram is collectively of lower individual entropy than Latin and Italian so substitution of one character in those initial two places likely will not increase the unpredictability.  But again, this is just a guess.