The stem look-up feature is very nice and efficient. But I'm sometimes embarrassed with the results.
For example, "inform" in the source segment is caught when "form" is in my glossary. Linguistically, this is correct because "form" is the root word of "inform" ... though I expected that, for the word "form", this feature would catch variations such as "forms" and "forming," but not "inform."
Well, I've got to be smarter than the Hunspell checker file to be able to make more productive use of CT.
Yep. I'm running against that wall too. Had a discussion about this with CafeTran's developer on Friday. He told me that the current implementation's just the beginning of a beautiful friendship.