Juliet & Shona:

LB has pointed out how you've got the wrong end of the stick about how
<interp> is to be used, but that aside, I think your query suggests you
may need to think a bit more about another issue.

Let's imagine we've extended the DTD with an element <gunge> so we can
reproduce your example without offending the parser.

You example would then look something like:

<gunge type="deep" value="not much">since no evil fate has bid thee come
on this road <gunge type="deeper" value="even less"> (for it lies far
outside the beaten track of men</gunge>), but right and justice. 'Tis
necessary for thee to learn all things, both the abiding essence of
persuasive truth, and men's opinions in which rests no true
belief.</gunge>

You say:

> we are concerned that in instances where
> multiple tags are nested, it will be difficult
> to determine which tag ends where.

and later ask (with the element name adjusted):

> While it is possible to figure out the nesting,
> is there any way of making the distinction between
> different </gunge> tags more explicit?

You need to consider here who or what is doing the "determining" or the
"figuring out". As far as an XML (or SGML) processor is concerned, there
is no difficulty whatever, in the example just cited, about determining
which element starts and ends where; and the distinction between the
different <gunge> tags (or, rather, different gunge elements) could not
be made more explicit.

So I assume you mean that human readers scanning your raw markup would
have these problems. That takes us to the heart of one of the most
fundamental and unfortunate misunderstandings about TEI markup among
humanities scholars, generally put less illuminatingly in complaints
that markup "gets in the way of" the "true" text. Yes, TEI markup is
meant to be human-comprehensible to those in the know, but TEI markup in
its raw form isn't meant for bedtime reading, or even routine scholarly
use. Raw markup is primarily for feeding to machines; and one of the
things you can ask machines to do with it is represent or re-write it in
ways that illuminate whatever your current (human) purposes are. Some of
the aspects of markup that humans find most tiresome when presented to
the naked eye are there precisely so that machines can reliably and
infallibly identify the text structure and make it available to us in a
more human-friendly form.

So, once you've sorted out the correct notation for the markup you want,
your problem becomes not "how can we spot the information we've encoded
into our markup by looking at the raw text?" but "what systems are
available for visualising the encoded information in ways that suit
us?".
The central concern of the TEI is how best to get the encoding in place.
The question of how best to make the information commodiously
retrievable is a different area, though obviously a related and
essential one.

Michael
---------------------------------------------------------
Michael Beddow   http://www.mbeddow.net/
XML and the Humanities page:  http://xml.lexilog.org.uk/
---------------------------------------------------------