the comparison of sh and ch

[Bluetoes101]

Do you have an opinion on this?
I'm not sure whether this topic is known with regard to (I call it) "mutations" in general. It is noticeable that the ‘basic words’ if we assume that the letters are prefixes (or whatever without belonging to the base word) are subject to a vowel swap from "o" to "a" or vice versa. Not all words, but too many to ignore this phenomenon.
Are there any theories or topics on this? Is there a language or dialect from the Indo-European branch that has such a vowel exchange?
I'll add a few examples

dor dar
dol dal
kooiin koaiin
chor char
chol chal
kol kal
kor kar
lol lal
dam dom

I posted a theory which I am still working on refining including "vowel" type things in the voynich script. I class these as EVA "aoy" + sometimes "e" (and "e" more often in Currier B)
Without getting too far into it to derail the topic, "a" and "o" seem to be somewhat interchangeable but if we have "a" not "o" we sort of know what's coming next.  
All of the examples you give I would classify as "modification" end to "blackslash" start transitions (some sort of flourish to a \ base shape glyph). You can continue this thinking to predict "what's next" pretty well, though meaning wise.. no idea. 

To give an example of this and the slight differences between "o" and "a" using made up "voynichese" words, imagine this is "guess the letter" game
1. sa_or
2. so_or

If I were playing the game I would exclude (for "1.")
ecgbshxutkpdfqzoay

And consider 
inrjmlv

I would then say
n,m are "word end" so no
j,v are weird so no
i, "ior" isn't really a thing
so "l" or "r".. we have an "r" already so I'll go "l". 

"2." is a bit more tricky, we have to do the same but also consider 
tkpdfqz

q is word start, and z is weird, so get rid
So I have tkpdf, and lr (from last guess logic) left to guess with

In this way "o" is a bit more difficult to predict, but "o" and "a" in the examples you gave both seem to be viable choices and used often. What that means.. ¯\_(ツ)_/¯