Desmond, you wrote:

> I don't see what's wrong with encoding a concrete feature of a
> manuscript with a concrete element.

Instead of "concrete" I would use the word "real".
Pagebreaks are real, and elements are real.  But reality
consists of two kinds of things.  Roy Bhaskar, in his book
"Dialectic: the pulse of freedom" (Verso 1993) calls them
"onts" (presences) and "de-onts" (absences).  Bhaskar claims
that one of the main failings of western philosophy is that
it does not recognize that absences are real and irreducible
to presences.  (Tobin, are you on this list? Can you help me
with this?)


> Using a milestone-like element such as <pb/> in fact is
> the only way to cleanly introduce the concept of pages
> into a manuscript composed of a sequence of other
> elements.

I don't think it is a *clean* solution.  There are many
things that can go wrong with it.  For instance: the new
page has the beginning of a new chapter.  Where do you put
your <pb/> element?  Inside the chapter heading just before
the first word on the new page, or just after the last word
on the preceding page, or somewhere in-between?  Or say you
have footnotes spanning several pages.  Suddenly there is
not just one page break but several.  Or what if the next
page is an illustration and the text continues after this?
My thesis is that this un-cleanness cannot be avoided
because it has a deep systematic reason, namely, that an
attempt is made to represent a de-ont by an ont.

This is why I am backtracking and asking: are we really
interested in the pagebreak itself?  Or are we more
interested in knowing which word is on which page?  This is
a cleaner approach because now we are talking about onts
only and have gotten rid of the de-onts.

> You appear to be suggesting that each and every word on a page be
> enclosed within another element, such as <page n="2"> fred</page>. I
> don't think many people could stand this.

Yes, each an every word *can* be marked up as you suggest, just
as each and every word *can* be marked up as <w>fred</w>.
What is wrong with this?

> and what about the frequent case where words are split by a page-break?

I guess in this case it depends on the philosophy of the markup.
If you consider words as atoms in your markup, then you would
probably say that the split word is on both pages.

> We would, following your scheme, have to enclose each
> letter with such an element or give each letter such an
> attribute. I don't see from what you say below how you can
> sensibly restrict this to only some of the words or
> letters on a page.

I wanted to leave it undetermined.  You don't *have* to mark
up every word, and you are not *prohibited* from marking up
any word that is on a given page.  For some purposes it
seems best to mark up that word which is closest to the
center of the page.  This undeterminenedness seems a natural
thing to me because in reality, many things are undetermined
and open too.  (Another illustration of the importance of
absences.)

Hans.

--
Hans G. Ehrbar   http://www.econ.utah.edu/ehrbar   [log in to unmask]
Economics Department, University of Utah     (801) 581 7797 (my office)
1645 Campus Center Dr., Rm 308               (801) 581 7481 (econ office)
Salt Lake City    UT 84112-9300              (801) 585 5649 (FAX)