Re: Topic and non-topic pronouns

Dans un courrier dat=E9 du 27/03/99 23:49:42  , vous avez =E9crit :

<<
 Does anybody know of any natlangs that use different third-person pronoun=
s
 (and/or different pronominal agreement markers on verbs) to represent top=
ics
 and non-topics?

 Here's an example of what I have in mind.  I'm assuming a verb morphology
 that has the following order of morpheme slots: subject - tense - object =
-
 stem (sort of like Swahili but without the noun classes).  I'll abbreviat=
e
 the subject and object agreement affixes as 3sT (third-person singular
 topic) and 3sN (third-person singular non-topic).  In both of the followi=
ng
 two examples, "the man" is the topic.

 man 3sT-Past-3sN-see woman =3D "the man saw the woman"

 man 3sN-Past-3sT-see woman =3D "the woman saw the man" or "the man was se=
en by
 the woman"


 This sort of like the proximate vs. obviative distinction in the Algonqui=
an
 languages, but it's different in that it doesn't involve a chain-of-being
 hierarchy and direct/inverse marking on verbs.  It seems to me that this
 accomplishes the same thing as a direct/inverse system (distinguishing
 between the core arguments of a transitive verb without the need for eith=
er
 case marking or fixed word order) with a lot less complexity.

 I've already thought of ways to extend this system to include focus marki=
ng,
 relative clauses, complement clauses, and indirect discourse, but they al=
l
 depend on the initial premise that this kind of pronoun system is plausib=
le.
 Since I don't know of any natlangs that do this, I'm afraid that there ma=
y
 be some deep language-universal reason why it's not plausible.
  >>

I think like you do about the 'unnatural nature' of this system. Only exam=
ples
I'm thinking of in natlangs are intransitive anti-passive and reflexive fo=
rms
:

'the cars sell well in Japan (*to customers)'.
'le poisson se mange le vendredi' (*par les gens)'

Most of my conlangs work like that :
(cases =3D a, i, u)
man A-see-I woman
the man sees the woman

man I-see-A man
the man is seen by the woman

me U[ man I-see-A woman]
I show the man to the woman

where A shows that the preceding actor is agent and I the next actor is
patient, and U temptative factitive ('quasi-ergative' ;-). Topic is either=
 one
put first.

But I don't know any natlang working like that, as if volitive voice, case
role and topic are always linked somehow.

Charles' Tomato conlang works like that too :

man goto-V1 woman =3D 'the man goes-to the woman'
woman goto-V2 man =3D 'the woman attracts the man'

where the volitive is always suggested in the first actor (hence 'topic' ?=
).

Mathias