A-reflexives (in 'subject' position)

[Wolfgang Schulze]

I guess that in many such instances of reflexives in pseudo-A or demoted

A-function the reflexive has a referential background at least in a
diachronic perspective (just as in Georgian, cf. Georgian _tavi_ <
'head'). This
would explain a) the possibility to have the reflexive in A function
without
on antecedent, and b) why this process seems somewhat related to
techniques of 'de-controlization':

Let us assume that the relationship between what now is a 'reflexive'
and its 'antecedent' once was a part-whole or a possessum-possessor
relation ('head > human being' in the Georgian case). Canonically we
would expect that the whole controls the part and the possessor its
possessum. Or, in other word, the 'whole' or the possessor would have
protoypical A-functions, whereas the 'part' or the 'possessum' is
related to O-functions. The inference 'possessor > A' vs. 'possessum >
O' is a well-known case in a number of language systems (esp. in those
that know a parallel 'ergative behavior' of their case or agreement
system. Let us put this into  a scheme:

        'whole' > antecedent        'part' > reflexive
        Possessor                        possessum
        Construed as A               construed as O
        Controler                        controlled

In case the roles switch, that is in case 'part' or 'possessum' etc. is
morphologcally or syntactically treated as A whereas 'whole' or
'possessor' is associated with O we arrive at a 'clash': A rather 'light

noun' (with reference to A) becomes A, and a rather 'heavy noun' (with
reference to A) becomes O. The inferential dynamics of such a clash are
self-evident: 'part' as A (which then becomes the 'head-less' reflexive)

acquires certain A-properties without matching them semantically. As a
result, 'part' or 'possessum' as A has reached a (rather) limited degree

of control, whereas 'whole' (or 'possessor') - which is atypical in
O-function if 'part' is present - looses at least certain 'portions' of
is controlhood, cf.

    'part' in A [-control -> + control]
    'whole' in O [+control -> -control]

Note that a prerogative of this analysis is the existence of a
referential background of the reflexive in question. I know that this is

true for a number of reflexives especially in languages that have a
relatively strong ergative behavior. It would be good to know of a
counter-example, that is a language that knows the construction noticed
by Nino (let's call it 'postcedent A-reflexives') but the reflexive of
which cannot be reconstructed as or related to a referential noun
diachronically.

Wolfgang

--
*****************************
Prof. Dr. Wolfgang Schulze
Institut fuer Allgemeine und Indogermanische Sprachwissenschaft
Ludwig-Maximilians-Universitaet München
Geschwister-Scholl-Platz 1
D-80539 München
Tel.: +89-21805343 / Fax: +89-21805345
Email: W.Schulze at lrz.uni-muenchen.de
http://www.lrz-muenchen.de/~wschulze/
*****************************