case syncretisms etc.

[Jürgen Broschart]

Dear fellow typologists,
if you read Lehmann (as Martin Haspelmath) suggested), you will see that
he says that no one has ever done really crosslinguistic research with
respect to case developments, but that
the order will presumeably be more or less the way indicated. Since I
have tried to
do empirical reseach of the type (forthcoming in my Habil), I know
exactly why it
is so dreadfully difficult to make any judgments to this effect. What
some people call
dative will be called allative or benefactive or purposive by others,
and ergatives figure
under ablatives, instrumentals, genitives, or vice versa. Also, by
looking at the
common lexical origins for various cases in Africa in a survey by Heine,
I discovered
that virtually any case development is possible. Nevertheless, there
were some
pretty solid tendencies for similarities between various case forms in
about 50 paradigms
I looked at: ESSIVE-COMITATIVE, COMITATIVE-BENEFACTIVE, ERGATIVE-
INSTRUMENTAL, INSTRUMENTAL-ABLATIVE, ERGATIVE-GENITIVE,
GENITIVE-ACCUSATIVE, ACCUSATIVE-NOMINATIVE,
DATIVE-ABLATIVE-LOCATIVE, SUPERESSIVE-SUBLATIVE etc.
All in all, there is a tendency for cases of similar status of
grammaticalization to show some overlap (nominative-accusative,
ergative-instrumental, superessive-sublative).
Secondly, there seems to be a very clear tendency for the development of
cases
which I had posited on theoretical grounds in Broschart 1994
(Praepositionen im
Tonganischen: Zur Varianz und Invarianz des Adpositionsbegriffs;
Bochum:Brockmeyer).
Overt cases start with meanings which are most different from the
'default' expecta-
tion. The default is a relation which is IDENTICAL with the centre of
the predication, PRESENT to the centre,  in a CONTROLLED relation, and
DYNAMIC. The 'marked'
cases are first NON-DYNAMIC (stationary), NON-CONTROLLED, NON-PRESENT
(non-contact), and NON-IDENTICAL. 'Normally', the defaults
are 'expressed' by zero, as in a canonical bitransitive relation like
'Peter gave Mary a
box', but in marked contexts (e.g. nominalisations) you often get
something like
'the giving AS the giving OF the box BY Peter TO Mary', where AS signals
identity,
BY signals presence or concomitance, OF signals control/possession of
the act, and
TO signals goal motion (dynamics). Looking  at these 'cases' (AS, BY,
OF, TO
or ESSIVE, INSTR, GEN, ALL) and their
historical origins, TO is still quite close to location (perhaps
stationary location as in
the German etymon ZU in 'Universitaet zu Koeln'), OF is recognisably
related to spatial
OFF (like most genitives), BY has lost a lot of spatial meaning, and AS
is least spatial
today (but it may go back to locatives; as far as I know, Japanese uses
the locative DE
also for essive meanings). My hypothesis is that grammaticalisation
starts with plain
'circumstantial'
locatives (either a general locative also including distant location
('off') or locative and
ablative), then the item acquires the feature 'motion' (as German ZU in
most readings),
then the relation becomes a control relation (as for OFF > OF), and
finally it also signals
concomitance (Latin ABL > COM/INS) and eventually identity in secondary
predications
('essives'). This development is unidirectional. I do not know any
convincing evidence
for the development of ESSIVE >LOC and INS > LOC, GEN>LOC, ALL > LOC,
rather than the other way around.
As I said above, the 'marked' cases will be anti-defaults,
which first tend to signal something which is NOT dynamic, NOT
controlled,
NOT concomitant, and NOT identical with the centre of the predication
(e.g. 'off').
Conversely, the typical default is not marked at all (predicates usually
do not have cases,
nor the immediate 'companion' of the predicate, the subject, nor the
controlled
object, nor the dative (usually goal of dynamic motion).
In English, the only remaining morphological case is the
'circumstantial' genitive, which is - on the other hand - the most
grammaticalized
case among the obliques (and it is historically usually a direct
development from an
OFF-locative or a general locative).
To sum it up: What gets marked first is what is most different from what
naturally belongs to an activity (identity with the activity, presence
to the activity, control over the activity,
and dynamics). Under certain 'marked' situations, the default may become
marked
(AS identity, BY concomitance, etc.), while the previously marked (e.g.
OFF) becomes the 'centre' of
the construction in marked circumstances (the 'genitive' OF). In other
words,
there is default-reversal (the periphery becomes central and vice
versa).
Incidentally, because of similar meanings, the ESSIVE (for secondary
predications)
may typically take over
'nominative' functions on primary predicates, and the INS (which
typically either comes from locative or
ablative) may finally turn into ERG, which may become
a subject nominative as in Easter Island.
Anybody (not) happy with this? Comments welcome!
---
Dr. Juergen Broschart
Institut fuer Sprachwissenschaft
Universitaet zu Koeln
D-50923 Koeln
am004 at rs1.rrz.uni-koeln.de
Tel. +49-221-470-2323