Knənʔtəəʔ: Ergativity and Relative clauses

So last night I realized that Knənʔtəəʔ is sort of ergative and I decided to go with that. So I needed to redo its relative clauses. And uh, I ended up getting exactly what I started with. Anyway, since relative clauses are interesting, I'll post my final explanation, where I basically justify what I already have. It's not all of my internal and external dialogue trying to figure it out, but it is something

"So I think I've figured out what was going on. There's two competing rules that I did realize I had going on until some analysis...Or maybe none of this makes sense and more research is need

anguishes

Let's analyze sentences together and see what happens:
bmis [so kaalbaa] dɨwõ kwiis
bmis so=k<aa>lbaa d<ɨ>wõ kwiis
man 3S=<REL>be.clean <CONT>eat seaweed
"the clean man is eating seaweed"
So we have a head which is the absolutive argument of the relative clause (which if not relativized would be kɨlbaa bmis "the man is clean". Since the relative clause technically takes a pronoun as its head, it works on a nom-acc alignment and fills the slot that would be occupied by the ergative pronoun, resulting in the pronoun preceding the relativized verb and everything being dandy. Let's work up to the next level of relative clause: non-stative intransitive verbs.

bmis [so lutɨhut] dwõ kwiis
bmis so=l<ut><ɨ>hut d<Ø>wõ kwiis
man 3S=<REL><PROG>sleep <PERF>eat seaweed
"the sleeping man has eaten seaweed"
Once again, the head is the absolutive argument of the relative clause, which free would be lɨhut bmis "the man is sleeping". Since the relative clause takes the pronoun as the head, it once again converts to nom-acc and fills the ERG slot(edited)

So far, so good. I've been maintaining the alignment (more or less :p) that I discovered yesterday

But then things get rough

"the man who ate seaweed is sleeping"
The relative clause is headed by a pronoun, so it goes nom-acc and should look something like :
Lɨhut bmis [so dwõ kwiis] or (Sï̵) bmis [so dwõ kwiis] lɨhut so (topic marking way to do it)
l<ɨ>hut bmis so=dwõ kwiis and (sï̵) bmis so=dwõ kwiis l<ɨ>hut so
<CONT>sleep man 3S=eat seaweed or TOP man 3S=eat seaweed <CONT>sleep 3S

Anyway, the restrictions with the way pronouns work basically mean (since agent incorporation isn't allowed and there are no voices) that relativization like this is limited to subjects=agent, despite the ergativity in main clauses. I guess technically these transitive ones could be interpreted either way, but that is abnormal for the speakers. However, part of the ergativity means that the main interpretation of a possessed verbal noun is that the possessor is the patient which allowed for a nice work around of which there are many ways to do it, based on topicality:

Kɨlbaa kwiis dnwõ so räp bmis or Sï̵ kwiis dnwõ so räp bmis kɨlbaa so so on and so forth k<ɨ>lbaa kwiis d<nØ>wõ so räp bmis
<CONT>be.clean seaweed <NOM>eat from man
"the seaweed the man ate is clean"
Notice that the possessor has to be removed from the possessive phrase to make the argument of the main verb clear. Other wise it could be translated as "The man's eating of the seaweed was clean" or "the man cleanly ate the seaweed"

Well, there you have it, something I should turn into a blog post. And it was all for nothing since it ended up looking the same as what I had originally anyway"

As for other things with ergativity, I plan on having deleted arguments follow an ergative pivot (basically "I hit the man and died" would mean that the man died, not that I died).