We discussed marking affixes with a ":"(5:prefix or suffix:5) and therefore stems can be identified as having no ":"
Line799 can then be changed, so it finds all stems in each parse of a word (and a word can have two stems if it is a compound):
Yes, and we will have to review every use of stems in the code in case they are affected by the change. The first thing to do is probably to implement "labels" in parses.
for "serialize_stems_and_words"
We discussed marking affixes with a ":"(5:prefix or suffix:5) and therefore stems can be identified as having no ":" Line799 can then be changed, so it finds all stems in each parse of a word (and a word can have two stems if it is a compound):