For a while I toyed with adding an attribute to a number of
elements that were found to be acting in this way: elements
that recurred, and behaved similarly, but were distinguished
by some functional feature. Short of using feature structures,
or abusing one of the rendition attributes, there didn't seem a
ready way to do this. (I think I tested out @role for this purpose.)

Martin's example seems to be of just this sort. Other examples
(not always involving <p>) might be:

1. lines of a hymn. Some lines are required (invariable); others
may optionally be skipped and are marked as such, e.g. by being
printed in smaller type.  LLllLlLL.

2. speeches or parts of speeches. Some plays present a full
reading text of a play but mark some speeches (or lines, etc.)
as bits to be skipped during actual performance. (Come to think
of it, the Congressional Record would do well to do likewise:)

3. marking source of a text. Some books distinguish the source
or authoritativeness of parts of its running text by typographic
means, e.g., the fourth edition might print everything new to
this edition in one font, but print everything carried over from
a previous edition in another font, so if you just want to read
the earlier text, stick to the text in the appropriate font. Similarly
with text rebutting an opponent: some paragraphs marked as
points on which the author agrees with the opponent, others
as divergent. &c &c

4. marking function within an argument. E.g. alternating paragraphs
representing opposing points of view (point / counterpoint); or
question and response; or doctrine and elaboration; or statement
and defense of same.

I'm still not sure that @type is quite right for this (since it feels
as though p @type should distinguish types of paragraphs, not types of
content), but I think I agree that the use case is real, and is perhaps
quite widely distributed.


On Wed, Apr 12, 2017, at 14:33, Martin Mueller wrote:
> I don’t think so. I think of <argument>  as a summary of what follows. A
> Gnomon article might have the following sequence of paragraphs, where
> upper and lower case distinguish the different types:
> PPPpPppPPp etc
> I’m sure I’ve seen this elsewhere. The author sends the signal: “you can
> skip this paragraph, but it’s part of a continuous argument.” You could
> of course encode it, using @rend or @rendition. But a @type attribute
> might be a better way of articulating this discursive structure, and you
> could derive particular layout instructions from the structural encoding. 
> On 4/12/17, 12:07 PM, "TEI (Text Encoding Initiative) public discussion
> list on behalf of Martin Holmes" <[log in to unmask] on behalf of
> [log in to unmask]> wrote:
>     This rather sounds like it might be a job for <argument>:
>     <
>     >
>     Cheers,
>     Martin
>     On 2017-04-12 11:04 AM, Martin Mueller wrote:
>     > I just stumbled an earlier thread about a @type attribute for <p> , and
>     > there was a question about use cases. Here is one I can think of right
>     > away. In the days long ago, when I regularly thumbed the Classics review
>     > journal Gnomon, I always appreciated reviews that had a main argument in
>     > larger type and subsidiary stuff in smaller type.  So you could read the
>     > whole thing or skip the fine print. You could  model the fine print
>     > stuff as inline notes, but I think of notes as bits of text that sit
>     > explicitly outside the text stream, while the alternation of paragraphs
>     > in Gnomon stays within a sequential reading order.
>     >
>     >
>     >
Paul Schaffner  Digital Content & Collections
University of Michigan Libraries
[log in to unmask] |