[or aiin]

[MarcoP]

I made a brute-force search for asymmetric word pairs. These are the most asymmetric couples (the two numbers are the occurrences of A.B and the occurrences of B.A). or aiin is indeed the most remarkable case.
I attach a more extensive list sorted by this arbitrary relevance measure: #(A.B)*[(#A.B-#B.A)/(#A.B+#B.A)]

Thanks for that list.  It reminded me of something I had noticed a while back looking at individual pairs of words, which is that words beginning in ch or Sh have a tendency to precede rather than follow words beginning with q.

If you sort them all out:
....

Also noteworthy is that most of the exceptions involve the word beginning in q containing an a-ending.

I agree, Sam. That is an interesting regularity.
If one considers qo-words ending with 'y', the fact that the patter repeats is quite clear (there's a single exception in one of the couples which have only 3 occurrences: qoty chy). Many of this couples are so similar that I doubt they are all meaningfully different: their shapes and behaviors are so similar that I think they could be spelling (or morphological?) variations of a single phenomenon. 

chedy qokeey 19 4 12.39
shedy qokeey 13 2 9.53
sheedy qokeedy 9 0 8.99
shedy qokedy 21 9 8.39
shedy qokeedy 16 6 7.27
chedy qokey 8 1 6.22
shedy qokey 6 0 5.99
shey qokey 7 1 5.24
shey qokeey 7 1 5.24
shey qokeedy 5 0 4.99
shedy qotedy 5 0 4.99
chey qokeedy 8 2 4.79
shedy qoky 9 3 4.49
cheey qokeey 4 0 3.99
sheedy qokedy 5 1 3.33
shedy qoteedy 5 1 3.33
shey qoty 3 0 2.99
qoty chy 3 0 2.99
chey qoky 6 2 2.99
chedy qotedy 3 0 2.99


Koen's observations are also something to look into. Once again, Emma has spotted a line of investigation that can help us understand more of Voynichese!