A Possible Reading of "rapa (turnip)" on f102v2

[ololololo]

There aren't enough letters in the Voynichese alphabet to extend that method to all the letters in the Latin alphabet, so words cannot be respresented this way

You can associate several letters with a single symbol. There is nothing extraordinary about this substitution technique. Homophonic ciphers have been known since the early 15th century

However, that is the wrong way 'round. 

In homophonic ciphers, the plain text has a normal number of characters, and the cipher text a much higher number.

Here, the cipher text (Voynich alphabet) is the one with the low number of characters.

So, is there a possibility that the Voynich encoder mapped several plain text characters to the same cipher character?
That looks like a really bad idea.

But this is effectively what people are assuming when they translate a Voynich character (say l) sometimes to l, sometimes to m, etc. (invented example, but this sort of thing is done all the time, usually because simple substitution can't work).

This approach will not be difficult for the decrypter. If I encrypt APPLE as OLUIA, and the L character has the values P and H, I think you will understand that AHPLE is the wrong decryption.
We shouldn't forget that the author made VMS for a person/people who could read it, but not for us.