Assuming it's a verbose substitution cipher with some ambiguities (solved by the order of substitution), the first step would be to identify tokens. Some very low entropy combinations are obvious: "yu" is always followed by "n ", "y" always precedes "un", so "yun " (31 instances) is probably a token. Same for "nuy " (17 instances).
Trying to build a key that looks plausible without thinking too much about the alternatives, I quicky get more than 26 combinations (around 50 and probably missed a few) so maybe I tokenized too many combinations, or there are homophones, 2 per clear text letter maybe, or nulls.
Anyway, it seems difficult to crack without a good algorithm and a computer...
Trying to build a key that looks plausible without thinking too much about the alternatives, I quicky get more than 26 combinations (around 50 and probably missed a few) so maybe I tokenized too many combinations, or there are homophones, 2 per clear text letter maybe, or nulls.
Anyway, it seems difficult to crack without a good algorithm and a computer...