Case syncretisms?

[Östen Dahl]

I think the notion "syncretism" may lead us somewhat astray. It suggests
that we have a set of predefined cases which are normally distinguished but
sometimes may get the same expression. But this gives the wrong picture of
the evolution of case systems. A common path is that from dative to
accusative -- e.g. "a" in Spanish. But when this happens dative marking is
at least to begin with applied only to a subset of all direct objects in the
language, normally the animate or definite ones. So we are dealing with an
encroachment of dative marking into accusative territory, not with a merger
of dative and accusative. Later on, the marking may spread to all direct
objects. But at this point marking of indirect objects may already be
undergoing renewal (something like this seems to be happening in Persian, if
I am not mistaken). So syncretism in the proper sense may never arise.

One reason that grammaticalization paths are not obvious for case systems
may be that in those languages where cases are used for signalling central
syntactic functions such as subject and direct object, they are normally of
a rather venerable age. For instance, the nominative and accusative endings
of Indo-European go quite far back in time and their origin is not
transparent. My general feeling is that people who study case systems have
been less ready to think in terms of grammaticalization processes than has
been done for instance in the study of tense-aspect systems. Maybe this is
because there are preconceived views about case systems that prevent us from
seeing what is going on.

- Östen Dahl