Well, yes, possibly, though not, I think without some violence to the
conventional definition of <cl> and <phr>. However, given that the exact
significance of these terms is highly theory-dependent, I think it would help to
provide a simple set of rules about how they are inter-related. Something like this:
1. Use <w> for the basic lexical items or units of meaning. 
2. Use <c> for non-lexical tokens such as punctuation
3. Use <m> for constituents of <w>s which also carry meaning but are not
complete in themselves
4. Use <phr> for sequences of <w>s which have some shared syntactic function
5. Use <cl> for sequences of <w>s or <phr>s which are regareded as syntactically
complete 
None of this precludes using <seg> for other kinds of segmentation more
appropriate to the theory in use, of course. 


In message <alpine.OSX.1.10.0808061034550.73177@lister.ei.Virginia.EDU> David
Sewell <dsewell@virginia.edu> writes:
> On Wed, 6 Aug 2008, Lou Burnard wrote:
> 
> > I think it might be better to partition model.segLike into those which
> > can reasonably be expected within a word (<m> and (possibly) <c>) and
> > those which cannot (<cl>, <phr>), I'm unsure about <seg>.
> 
> If one were using TEI to mark up text in an agglutinative language,
> wouldn't there be cases where <phr> and possibly even <cl> could
> legitimately appear within a <w> tag? Maybe this is one of those cases
> where we need to provide maximum flexibility in content models even if
> it allows for valid but nonsensical tagging? (Seems to me that anyone
> who is serious enough to want to use the linguistic segmentation tags
> will likely be capable of analyzing and tagging grammatical structure as
> needed.)
> 
> David
> 
> -- 
> David Sewell, Editorial and Technical Manager
> ROTUNDA, The University of Virginia Press
> PO Box 801079, Charlottesville, VA 22904-4318 USA
> Courier: 310 Old Ivy Way, Suite 302, Charlottesville VA 22903
> Email: dsewell@virginia.edu   Tel: +1 434 924 9973
> Web: http://rotunda.upress.virginia.edu/