correspondence encoding problems: letter headings

Ron Van den Branden <[email protected]> Wed, 22 Feb 2017 15:54:56 +0100
Newsgroups	gmane.text.tei.general
Message-ID	<[email protected]>
Hi,

We're (finally) moving to TEI P5 for correspondence encoding, and would 
like to stick to "vanilla" TEI as much as possible. While the TEI 
correspDesc module does a great job in defining and explaining elements 
for encoding metadata for letters, we often find ourselves struggling 
with ways to transcribe actual text elements.

One is the letter heading. At first glance, something like 
(ab|div)[@type='letterhead'] would seem a plausible candidate encoding 
(I've found a concrete example on slide 40 of the presentation at 
https://drive.google.com/file/d/0B_1qUxvG29kvekJDZzQzZ1JZT0E/view). Yet, 
it seems to be more complicated than this. Typically (but not 
necessarily), letter headings occur at the start of a letter, before the 
opening salutation, etc. However, <div> nor <ab> are allowed before 
<opener> (or any other model.divTop elements). Following encoding hence 
is invalid:

   <!-- invalid -->
   <body>
     <div type="letter">
       <ab type="letterhead">...</ab>
       <opener>...</opener>
       <p>...</p>
     </div>
   </body>

Looking for possible (TEI) solutions, these seem to be the options:
     [1] abstraction: encode <ab type="letterhead"> after <opener>
         [+] easy; <ab> is flexible enough to occur in many places 
(except before and inside <opener>)
         [-] less faithful to the source
     [2] encode <seg type="letterhead"> inside <opener> (since <div> nor 
<ab> are allowed inside <opener>)
         [-] not sure if this could be considered as part of the opener, 
really; <seg> is an inline element
     [3] encode letter heading as <opener rend="letterhead">, sibling to 
the regular <opener>
         [+] quite elegant (and <closer rend="letterhead"> would be a 
viable analogy for letter headings occurring at the end of a letter)
         [-] tag abuse?  what about letter headings occurring at 
different locations?
     [4] encode letter heading before <opener> as <head 
type="letterhead">, *if* it can be considered a title to the letter, really
         [+] could be useful without too much "semantic stretch" for the 
<head> element
         [-] <head> can only occur as the first element, so unusable for 
encoding letter headings at different locations
     [5] encode letter heading before <opener> as <figure 
type="letterhead">, *if* it can be considered a graphical element, really
         [+] <figure> is flexible enough to occur in many places (even 
inside <opener>, if needed), and leaves different encoding options:
                 -only signal its presence: empty <figure 
type="letterhead"/>, possibly with a description in <figDesc>
                 -transcribe its textual contents inside <figure>
                 -include a digital image with <graphic/>
         [-] tag abuse?

Following example illustrates these options in different <div 
type="letter"> sections:

     <body>
       <!-- option [1]: ab[@type="letterhead"] after opener -->
       <div type="letter">
         <opener>...</opener>
         <ab type="letterhead">...</ab>
         <p>...</p>
       </div>
       <!-- option [2]: seg[@type="letterhead"] inside opener -->
       <div type="letter">
         <opener>
           <seg type="letterhead">...</seg>
         </opener>
         <p>...</p>
       </div>
       <!-- option [3]: opener[@rend="letterhead"] before actual opener -->
       <div type="letter">
         <opener rend="letterhead">...</opener>
         <opener>...</opener>
         <p>...</p>
       </div>
       <!-- option [4]: head[@type="letterhead"] before actual opener -->
       <div type="letter">
         <head type="letterhead">...</head>
         <opener>...</opener>
         <p>...</p>
       </div>
       <!-- option [5]: figure[@type="letterhead"] before actual opener -->
       <div type="letter">
         <figure type="letterhead">
           <figDesc><!-- a description --></figDesc>
           <graphic url=""><!-- a link to a graphical representation --></graphic>
           <!-- transcription of text contents -->
         </figure>
         <opener rend="letterhead"><!-- option [3] --></opener>
         <opener>
           <seg type="letterhead"><!-- option [2] --></seg>
         </opener>
         <ab type="letterhead"><!-- option [1] --></ab>
         <p>...</p>
       </div>
     </body>

Am I overlooking any other options? How is this being dealt with in 
other projects?

Personally, I'm getting very much in favour of option [5] (<figure 
type="letterhead">), since it
     -is flexible (encoders can choose how "deep" they want to encode 
the letter heading)
     -allows for uniform encoding of letter headings (<figure> can occur 
anywhere)
     -is quite faithful (letter headings do have graphical qualities, 
after all)
Or are there reasons not to take this option?

Many thanks for your thoughts,

Ron