Clumps and Runners

1
Clumps and Runners
Bamboo Workshop, Tucson AZ, January 13, 2009
John Unsworth
2
Exchange of ideas

• A substantial amount of the work of the
  humanities is carried out in the form of the
  exchange of ideas. -- Anthony Cascardi
• No ideas but in things -- William Carlos
  Williams

3
Things in the humanities

• the whole variety of material cultural heritage,
  from images to sound to text to maps to
  architecture, landscape, and all manner of
  carrier media ... including sheep DNA in
  parchment.

4
Representing things

• ... a problem usually discussed at a high level
  of abstraction, in digital humanities. I'd like
  to discuss it at a very prosaic level, instead,
  with text objects as the example.

5
TEI

• "The Text Encoding Initiative Consortium is an
  international organization whose mission is to
  develop and maintain guidelines for the digital
  encoding of literary and linguistic texts. The
  Consortium publishes the Text Encoding Initiative
  Guidelines for Electronic Text Encoding and
  Interchange an international and
  interdisciplinary standard that is widely used by
  libraries, museums, publishers, and individual
  scholars to represent all kinds of textual
  material for online research and teaching."
• http//www.tei-c.org/About/index.xml

6
Lessons of Google?

• But why do any mark-up at all? Isn't the lesson
  of Google that all that really matters is the
  word, and tagging is superfluous? Not if you want
  to select out of Google Books
• Fiction
• Books about England
• Books published in the 1900s
• Books written by women

7
But structural markup? Really?

• It turns out that, more often than not,
  paragraphs and chapters, lines and verses, are
  meaningful units of composition, and therefore
  they can be meaningful units of analysis. It can
  also be important to recognize the difference
  between the main text of a work (the chapters in
  a novel, say) and the paratext (table of
  contents, preface, running headers, etc.),
  especially if you're asking statistical questions
  about the text.

8
Surely we don't need to mess with the words,
though...

• ... at the word level, it can be helpful (in a
  novel, for example) to ignore proper names, for
  example, so as to see more clearly what's going
  on with other kinds of words--but even at the
  word level, the information about the words is
  carried, ultimately, in tagging.

9
Interoperability

• The TEI community believes. . . that "people use
  TEI in many different contexts for many different
  purposes to encode many different kinds of
  material."  But they also believe that this
  somehow, in some universe, achieves the TEI's
  stated goal of interoperability.  It really
  doesn't.  So if people are in fact encoding
  things in all sorts of different ways and for
  different purposes, then why shouldn't I chuck it
  all and roll my own?  You say that's it better
  not to go it alone.  
• (Steve Ramsay, in email)?

10
Interoperability?

• As we are coming to the end of this project --
  and returning to an earlier exchange of views on
  the Monk list about interoperable texts-- I can't
  refrain from pointing to the large amount of
  needless and heedless divergence. There is good
  and bad news about it. The bad news is that it
  has caused a lot of work. The good news is that a
  very high percentage of problems can be solved
  quite satisfactorily by supplementary conventions
  to the content rules of elements. If, for
  instance, the people who slapped the Level 4
  Guidelines together had spent two hours about
  making recommendations what to do or not to do
  about soft hyphens at the end of a line or page
  when you encode a text according to Level 4
  Guidelines, we'd have a lot less grief. And so it
  goes with a lot of other little stuff.
• (Martin Mueller, in email)?

11
Clumps and Runners

• So, we have data in clumpsin collections that
  are curated and hosted by libraries, publishers,
  and others what we need are the runners that
  connect those clumps, and what we'll discover
  when we have them is that data doesn't move
  between clumps very successfully. That's a
  problem that nobody is really dealing with, and
  that's a role for Bamboo. I'm not recommending
  that Bamboo become a standards group, but rather
  that Bamboo attend to the actual problems of data
  interchange and interoperability, in actual data
  domains, for actual research projects, and that
  it collect and synthesize the experience of
  actual practice and turn it back to the stewards
  of content, to reduce needless and heedless
  divergence.