From lojbab@access.digex.net  Sat Mar  6 22:45:19 2010
Date: Tue May 23 03:12:53 1995
From: Bob LeChevalier <lojbab@access.digex.net>
Subject:      Re: Reflexivity and {ri}
To: "Dylan P. Thurston" <dpt@abel.MATH.HARVARD.EDU>
X-From-Space-Date: Tue May 23 03:12:53 1995
X-From-Space-Address: lojbab@access.digex.net
Message-ID: <8sWBKqdtloH.A.n7E._t0kLB@chain.digitalkingdom.org>

Dylan:
>Actually, I'd recommend changing the behaviour of {ri}.  From the
>history I know, it seems like the current behaviour was necessary when
>it was the only form of pronoun; but now {ko'a}, {ra}, etc.  (not to
>mention lerfu) cover pronouns quite sufficiently.  Specifically, the
>antecedent of {ri} should be the sumti whose termination is closest on
>the left.

That is close to the default behavior.  However you skip most
monosyllable members of KOhA as being unnecessary to ri-ify.

But wait - what means termination closest?  Do you count elided
terminators or not?  If you think of a main complex sumti with relative
clauses and specified description sumti, "last terminated" might mean
the whole sumti takes precedence over the pieces.  This might be
desireable, but if you want one of those pieces it will be hard to get
to them in-mind.  On the other hand, such main sumti can usually be
accessed with vo'a-series, if complex enough, assigned a ko'a with
little increase in confusion since it already is complex, or most
clearly replaced by a BY or "le sumti" style anaphora.  Those tend to be
most clear when working at the gross levels of the sentence. ri and ra
are meant to work like popping a stack, and the last-terminated is
often not the obvious thing to pop up on the mental stack.

lojbab