Re: relative tense / universals / object agreement

Boudewijn Rempt wrote:

> I've only once made a conlang with shows object agreement, but in
> that case I wanted to do it too neatly, and provided different
> affixes for S, O and A, a situation that practically doesn't occur
> anywhere...
What do you think of so-called "active" systems where a subject of an
intransitive verb might be either an agent or a patient, and is marked
accordingly? To me, ergative/nominative systems both seem over-optimized.