Nominative to ergative shift

I just re-discovered an old project of mine for a
language derived from Drasele'q. D. is a nominative
language, but this derived lang has become ergative.
Is this likely to happen? I know that PIE is believed
to have been ergative, and then shifted to nominative.
What I'd like to know is if you have any idea if the
opposite could happen, or has happened.

Here's the process of change:

Drasele'q: VSO,
 Verb[ending] Subject[null_mark] (Object[Acc])

Stage 1: V(O)S, SVO (hesitating)
At this stage the word order changes, and the accusative
mark begins to disappear.

 Verb[ending] (Object) Subject
 Subject Verb[ending] (Object)

Stage 2: only V(O)S
The verb ending weakens and tends to become a "hanging
consonant", so it joins the following words (liaison).

 Verb [vb-end]Object Subject  (transitive sentence)
 Verb [vb-end]Subject         (intransitive sentence)

The previous verb ending becomes a full-fledged
proclitic to the next word in the sentence, so the
object of a transitive sentence is marked like the
subject of an intransitive sentence. Quod erat facendum(?).

Then of course there could be another word order shifts,
you could insert adjectives, etc., but the people had got
accustomed to most simple <verb noun noun> sentences with
these marks, and they kept them even when the order was not
the standard one.

The main weird thing about this is that the marked nouns
are "absolutive", which is usually *un*marked AFAIK in most
ergative languages. And it's a different mark according to
the person.

What do you think about this? Comments, please!


--Pablo Flores

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
 Alexander Graham Bell's Observation:
  When a body is immersed in water,
  the phone rings.