On Mon, Sep 09, 2002 at 10:21:53AM -0600, Dirk Elzinga wrote:

> >Karapcik, Mike <[log in to unmask]> writes:
> >
> >>          Japanese doesn't have articles. The dactyls kore/sore/are and
> >>  kono/sono/ano are sometimes used like a definite article, though they both
> >>  basically mean this/that/that over yonder.
> At the risk of getting lost in the crowd, here's a sketch of
> Shoshoni's rather complex but pretty nifty deixis system.


> si-, i-  'here'
> sai-, ai-  'here (but further away than si-, i-)'
> sa-, a-  'there (but in sight)'
> su-, u-  'there (out of sight)'

This brings up an interesting question: does anyone know of a theory as
to why deictics are often so very regular, for example with a prefix
indicating level of deixis and a suffix indicating the domain in which
it is applied, even in languages that are not that big on regularity?
English, which is so often irregular, still has some mostly-intact h/th/wh
triads (although it also has non-compliant forms like this/that/which, and
partial triads like now/then/when (was there ever a "hen"?)).

Are there strong regularizing forces at work in this area, back-forming
corrections to irregularities that (surely must) creep in?