Theory: perhaps the letters do not have phonemic content?

[ololololo]

Such a "semantic" system is too difficult to implement, read, and practice. This is incredibly expensive on a VMS scale.
Of course, if you're a beginner, this may not be possible, but I would like to see examples of words from the manuscript that could match or resemble your pattern.
As I understand it, you consider Voynichese to be a philosophical language. Well, this hypothesis is quite vague, but it's also interesting.
However, this idea may be feasible as a cipher, but not as a separate semantic system...

I will admit that the semantic aspect of the script is an unfounded idea in retrospect, and I agree it is definitely more feasible as a pure cipher/shorthand. A good test of the semantic theory would probably be trying to find patterns in, say, the "labels" in the Herbal section, but looking at them now, I don't think there is a correspondence between the letters and any category of thing one can find. "Semantic" letters are probably just a bust.

Of course, the first guess would be to look for a connection in the gallows, as they are often found at the beginning of plant "names." However, this will not provide us with any information, especially when we find plant names that do not begin with a gallows (okeodof). 
Gallows do not indicate a category or anything similar. They most likely perform the same function as the other symbols.