This is a description of this project
Most of my work errs on the more theoretical side, at least as far as linguistics is concerned, so this project was a particularly fun contrast.
Amharic has object markers, morphemes which attach to verbs and mark objects. So you wouldn’t expect to see them at all on intransitive verbs like “run” or “laugh”. But you do, and when you do, they come with a whole host of unexpected meanings, like dares, warnings, rhetorical questions and exclamations of affect.
For instance, a parent might put one of these object markers, in fact a feminine one, on “return” in “He’s going to return”, to mean “He better return, and he’s in trouble when he does”.
Like you’d expect, the meaning is a bit ineffable, and it took a while to pin down the semantics. It looks like the object marker (or at least the syncretic morpheme that morphologically behaves the same as the object marker) triggers a presupposition that the event of the run (e.g. returning home) is not optimal in some modal sense. The feminine object marker presupposes that the event is not optimal for the subject of the verb, while the masculine object marker presupposes non-optimality for the speaker of the sentence.
This is a pretty weird thing to happen, but does nicely unify the semantics and predict some distributional facts, like that the feminine never appears when the subject isn’t animate.