--- In [log in to unmask], Chris Bates 
<chris.maths_student@N...> wrote:
> >
> >Hi.
> >I just finished Anna Siewierska's "Person", and I think I can 
> >summarize what that book had to say about this question in re 
> >natlangs.
> >  
> >
> Cool. I've skimmed through that book, but unfortunately I don't own 
it.

I don't either.  I had to give it back.

> 
> >In other words:
> >In NatLangs, diachronically,
> >it is far more common for a person-marked pronoun to be lost,
> >and replaced by a 3rd-person noun,
> >than for a noun to become marked with person.
> >  
> >
> What basically happened in my conlang is that 3rd person pronouns 
> started being used as a kind of article on nouns to mark gender 
(there 
> is a gender system). This is kindof important because the language 
was 
> developing verbal agreement which included gender and there is no 
case 
> system, so... but anyway, this has happened in some natlangs so I 
don't 
> think there's really a problem. But I didn't think it was too much 
of a 
> stretch to also include in the system markers for 1st and 2nd 
person 
> referrents also, given that 3rd person pronouns started being 
> consistently used with nouns. So the motivation for the system was 
more 
> or less gender marking, and the person agreement was just an 
accidental 
> extention to the system.

Sounds plausible to me.

Anyway, as you can see from the many posts about Elamite and a few 
about Nahuatl, in spite of the clear [and unassailable] logic of 
[Siewierska's] accessibility-based argument that this kind of thing 
should be rare in natlangs, it does in fact happen.
[words in square brackets above are those which Siewierska might not 
agree with.]  

Tom H.C. in MI