Dan O'Donnell wrote:
> b) For Lou: in addition to what Martin said about the tool att, the idea
> came up I think in a discussion of how a creation app might be
> conceptually similar to langusage, which is itself a type of feature
> structure, I think.
I beg to differ.
A language, an application which creates something, a linguistic
structure, or indeed a kettle of old socks might all be *described* by
means of a feature structure. That is because a feature structure is a
generic method of representing interpretations as named feature-value
pairs organized in various ways. It does not mean that the things a fs
describes are *types* of feature structure.
I.e. a place for identifying properties that can be
> referred to elsewhere by ID:
This is a very restrictive definition of a feature structure! The TEI
feature structure representation *uses* the id/idref mechanism inherited
from SGML for economy of representation. It is by no means an essential
part of it.
> In this case, I think the creationapp proposal seems to lie between
> langusage and revisionDesc in syntax and function.
I don't know what dimension you are plotting these things along, but I
think they have very little to do with each other. One is about the
languages used in a text (and note that it is redundant for most
documents come P5). The other is about what happened in the process of
creating the document. Their syntax is identical -- they both use XML.
Their functions are as different as chalk and cheese.