case syncretisms, etc. (forward from S. Gildea)

[Scott DeLancey]

---------- Forwarded message ----------
Date: Fri, 23 Oct 1998 12:30:25 -0500
From: Spike Gildea <spikeg at owlnet.rice.edu>

re: case syncretisms, etc.

A short note --

The discussion to date has largely missed an important aspect of the
evolution of case-markers, namely that they often become polysemous not
through simple extension from one contruction to another, but that through
reanalysis of a construction which contains the case-marker in a prior form
being reanalyzed as another construction, in which each of the morphemes
(including the case-markers) participating in the construction gains
additional meaning/function.  In both cases, we have not simply the
evolution of the case-markers in question, but the evolutions of the
constructions in which the case-markers occur; in the case of reanalysis,
the connection between the prior and resultant meaning of the case-markers
is only indirectly related, as mediated by the functions of the entire
constructions.  For instance, when a nominalized verb plus its logical
arguments (in whichever cases those might be expressed) is reanalyzed as a
main clause, the possessor of the nominalization (i.e. genitive
case-marking) might become nominative, accusative, or absolutive, depending
on the morphosyntax of the source construction.  In these cases, the
semantic relationship between the "donor" and "recipient" functions of the
case-marker play little apparent role in motivating the reanalysis (which
makes it of some interest to explain why they nonetheless follow relatively
consistent typological patterns).

The difference between EXTENSION and REANALYSIS is very clearly discussed
in Harris & Campbell's 1994 book HISTORICAL SYNTAX IN CROSS-LINGUISTIC
PERSPECTIVE, and they give many examples of case-marking changes of both
types.  With apologies for self promotion, I also present several
case-studies of both extension and reanalysis in verbal person-marking
categories, as well as reanalysis in nominal case-marking patterns, in my
1998 book ON RECONSTRUCTING GRAMMAR: COMPARATIVE CARIBAN MORPHOSYNTAX.  In
the Cariban family alone we find  allative/dative to ergative, ablative to
ergative, genitive to ergative (then on to nominative), genitive to
accusative, and genitive to absolutive (all via reanalysis of entire
constructions), as well as a large number of extensions between A, Sa, So,
and O person-markers on verbs (in which the constructions as a whole
arguably change very little).

Spike