>
> The verb accords with its subject in IE; if there are more subjects
> than one, the verb is put in the dual or plural. The same is seen in
> the related families outside IE. Therefore, if this represents a
> change from a presumed older ergative structure, that change is
> itself very old indeed and so of questionable use in an analysis of
> the IE facts.
>
According to Dixon: 'Ergativity' the main way for ergative languages
to change into accusative languages is this:
intransitive sentences
(subject in the absolutive + verb)
are not changed
If the verb agrees with the subject, nothing's changed in that respect
either.
transitive sentences
(subject in the ergative + object in the absolutive + verb)
are gradually replaced by a competitor sentence type (the anti-passive)
(subject in the absolutive + object in some case like allative + verb)
Ex hypothesi the verb agreed with the subject in intransitive
sentences, so nothing's changed here either.
Now the accusative has some remaining allative functions in IE: Latin
Romam etc. Finnish has a to-case in -n: Helsingin "to Helsinki", also
used as accustive, I believe. The PIE accusative might therefore have
been a pre-PIE allative.
PIE nominative is part endingless, part has -s, most likely from the
deictic *so. The PIE nominative might therefore have been a pre-PIE
absolutive.
In other words, nothing stands in the way of PIE being descended from
an ergative language with absolutive in *-Ø, and allative in *-m, and
with the verb agreeing with the absolutive (what the verb agreed with
in the transitive sentence we will never know, since that sentence
type has gone extinct).
The same argument appplies to the FU languages, of course. Therefore,
the fact that they are accusative does not imply they always were,
specifically not all the way back to the split between them and IE.
Torsten