Case syncretisms?

[Michael Noonan]

As Oesten correctly points out, what is most interesting for typologists
is not simply syncretisms but rather paths of grammaticalization.  Juergen
Broschart mentions a number of such cases from his crosslinguistic
research.  I'll just mention a couple of cases here, similar to ones
discussed by Juergen, that are typical of the Himalayan speech area, the
subject of a number of recent postings.

Perhaps the most widely attested development is:

	LOCATIVE > ALLATIVE > DATIVE > ANTIDATIVE > ACCUSATIVE

The Tibeto-Burman etymon la/ra can be attested at every stage of this
development.  Bits of this grammaticalization chain are, of course, widely
attested elsewhere. Matthew Dryer, Ivo Sanchez, and I have discussed these
developments in various places.

Also common in this region is:

	INST > ERG/INST > ERG/INST/ABL

The line of development usually follows the following path:  the INST
extends to ERG senses, sometimes, as in some Newari dialects, replacing an
earlier ERG, and then the new ERG/INST either extends to include ABL
senses [means > agency > source] or constitutes a component of a
compounded case form signaling ABL. An example of a compound ABL is found
in Dolakha Newari -lan, which Carol Genetti speculates derives from an old
ABL -la plus ERG/INST -na/-n. A similar case is the Chantyal ABL -gAm-sA,
consisting of -gAm, which may independently signal ABL, but which is
usually found compounded with ERG/INST -sA.  The compound is often reduced
to -gAN-sA [N = velar nasal] and -N-sA. The other Tamangic languages [save
Gurung] have an ERG/INST/ABL cognate with Chantyal -sA and it's a moot
point whether this developed from an ERG/INST via extension or via
compounding cum phonological reduction: it would be a short step from
Chantyal -N-sA to -sA. None of the other Tamangic languages shows [so far
as I know] any trace of -gAm, however.
	Another line of development for ABL takes a LOC and compounds it
with an ERG/INST.  Pattani illustrates this, wherein LOC -iN compounds
with ERG/INST -zi to form ABL -iN-zi.  [Chantyal -gAm may be another
example of this, since this ABL is probably in origin a compound
containing a locative -kV.]
	In any case -- and a propos of the original question which sparked
all of this discussion -- while ERG/INST and ERG/INST/ABL conflations are
commonly attested in the Himalayas, ERG/ABL conflations excluding INST are
not, and there seems to be a principled basis for this.

Mickey

Michael Noonan			Professor of Linguistics
Chair				Office:   414-229-4539
Dept. of English	       	Fax:	  414-229-2643
University of Wisconsin	       	Messages: 414-229-4511
Milwaukee, WI  53201		Webpage:  http://www.uwm.edu/~noonan
USA