Peter_Constable@... wrote:

>
> On 11/11/2001 11:45:38 AM Michael Everson wrote:
>
> >This is historical and it is interesting.
> >
> >What would a classification based on "script properties" entail, and
> >why would it be interesting?
>
> I agree that the history is interesting because it tells us how things
> came to be, and it tells us how things can change to be other types of
> things.
>
> A classification based on script properties is interesting because it
> helps us understand what they can potentially be like and how they work.
> It helps people who are trying to learn about the full range of varieties

But there _isn't_ any limit on the range of varieties or what they can
potentially be like, because they are the products of human imagination
-- as opposed to languages, which we have to believe are constrained,
somehow, by our inborn linguistic ability. There's no logical reason
that Ross-constraints should hold, but they do.

As I keep having to point out, writing is not language. (See list of
ways in Blackwell Handbook of Linguistics.)

> to begin understanding from a small set of prototypical examples that
> covers a majority and then work gradually out to the less typical
> varieties. It may also help people involved in trying to develop certain
> kinds of automated processing to divide the potential problems into
> different subtypes that may have their own issues. For example, if someone
> is developing text-to-speech processes, they may want to develop different
> varieties of algorithms depending on the nature of the relationship
> between structural elements and the linguistic objects being represented.
>
> I'm sure others might be able to think of other reasons why a
> classification based on script properties is interesting.
>
> What does it entail? Defining an explicit set of classifying criteria
> based on a consistent set of principles (e.g. all classes are defined in
> terms of the same set of properties) that are clearly applicable to
> prototypical examples and that bring to light patterns that someone finds
> interesting (for whatever purposes).
>
> And, no, it does not entail that every object must fit cleanly into
> exactly one class without any fuzzy boundaries. These things are defined
> with respect to prototypes. The objects don't have to be free of fuzzy
> boundaries, but the yardstick has to be well-defined.

--
Peter T. Daniels grammatim@...