Detail #404: A Really Tiny Remnant of an Ergative System

The normal situation in an ergative language is that the absolutive case is the unmarked case, and one would expect a transition to nominative-absolutive to keep the absolutive as nominative.

However, in nom-acc languages, a grammar change that sometimes happens is "accusativism", the replacement of the nominative by the accusative form. We can imagine that a similar thing could happen with ergative turning into a nominative, and some other case replacing the accusative (or not at all).

For personal pronouns, it seems even less peculiar for something like that to happen - maybe the ergative and absolutive are suppletive anyways, and further case forms are formed by further suffixes, which could muddy the waters with regards to which form is more marked in the first place. So, after all that handwaving, let's posit this for the third person pronouns:

ERG -> NOM,
DAT -> ACC (or maybe ERG -> ACC)

A situation with ERG -> NOM, ABS -> ACC is not entirely impossible, and would enable what I am going for here, but I find it typologically fairly unlikely. Also, I imagine this idea would also work in an Iranian-style split ergativity.

Now, for the tiny remnant. Let's imagine the language requires dummy subjects sometime. Let's imagine that in this particular context, the old third person inanimate absolutive survives.