Keeping Track of Ambiguity in your Conlang?

Has anybody come up with a way of tracking accidental homonymy (due to
inflections or compounding) in their conlang, short of devising an
unambiguous scheme for its morphology?

Making it guaranteed-unambiguous only works for really regular agglutinating
or isolating things, and I've always wanted to do something with lots of
fusion and internal inflection.  I could accept a certain amount of ambiguity
as naturalistic, but I'd want to keep it limited to a reasonable amount, as
well as come up with likely paraphrases that the speakers would use in place
of overly-ambiguous words.

So I'd like to write something that would take the lexicon and apply
every valid transformation to every word, recursive to several levels,
and then find duplicates.  Has anyone tried this before?

Amanda

Replies