Re: Evidentials

On Sat, 03 Sep 2005 01:05:10 -0400, John Quijada <jq_ithkuil@...>
wrote:

> Paul Bennett wrote:
>
>> I'm thinking about bringing optional evidentials into Br'ga
>> inference
>> What other forms and functions should I be considering?
>> =========================================================================
> You might distinguish whether a hearsay source is considered 1)
> trustworthy, 2) unknown if trustworthy or not, 3) untrustworthy based on
> innocent error with no intent to deceive, 4) untrustworthy based on an
> intent to deceive
Reasons for trust, as well as extent of trust? I can picture at least a
tree structure forming -- equally modellable by a grid with a diagonal
half blacked out, but at least it's a start.

I can probably do something in the "I examined the
Agent/Patient/Tool/Benefactor but I suspect/know it to be
misleading/faulty/forged" arena, too.

Yikes. I think it's a very good job that these structures will be
optional. Things already get reasonably hairy with such horrific
structures as the pro-anything and related underspecification
possibilities.

I think I may have to rethink the whole dealio. Maybe.

Thanks, I think,



Paul