Theory that the final section is a multilingual glossary

[eggyk]

Adjusted EVA: k = tl,     t = thl,    l = th,  y = -us / con-, m = ré / ch = er / sh = ér

If these are the right phonetic mappings, the underlying text has no K, L, M sounds at all, which seems very strange. I'm not sure I know a language with wouldn't have at least one of these three in some form.

So these mappings aren't intended to be a correct phonetic mapping, not at all. Those changes are not an extensive list of possible changes that could be done, nor an attempt at a genuine substitution. To an extent they are arbitrary. 

The point is that mapping similar looking symbols to similar sounding sounds totally changes which words correlate with which and in my view shows clusters of connected words that you may expect from such a glossary.

For example, "qokeedar otedy qokeedy" dont look all too similar, but after those changes you have 3 very similar words "qotleedar othledus qotleedus".