This Thing of Darkness (Part III)

To finish this trio of posts, I would like to provide a response to Mike Witmore’s post at Wine Dark Sea labeled Shakespeare Out of Place?.  In this post Witmore notes that “when we did a further study of the clusters containing works by Shakespeare, we noticed that their elevated levels of two different LATs that dealt with punctuation – TimeDate and LanguageReference – was an artifact of hand modernization.”  These two Language Attribute Types (LATs) lie within the Docuscope’s Dimensions of Narrative Time and Reference Language and Docuscope’s Clusters of Narrative and Special Referencing respectively.  Witmore notes the difference between the clustering of Shakespeare’s plays among their contemporaries before and after the emendation of these LATs.  I must presume that Witmore also chose to use Ward’s Test at the LAT level of Docuscope’s output in JMP to create his diagrams, since he doesn’t mention adjusting any of the Dimensions or Clusters so therefore wouldn’t have changed diagrams.  I created similar diagrams to reflect on the diagrams I created when first starting out, for my post Epic.  For that post, I made a diagram looking solely at Shakespeare’s corpus in the First Folio.  I then recreated the experiment a second time, but left out the two LATs mentioned by Witmore above.  Both portraits are side by side for comparison below.

I chose to highlight texts that “moved”, or at least aren’t in the same order after removing the two LATs in the second picture.  As you can tell, transitioning from ninety-six to ninety-four LATs can really change a diagram’s appearance and structure.  Witmore as well, notes this large scale change although he also states that “by eliminating these, we lost their sorting power on the rest of the corpus, so there was a tradeoff.”  The tradeoff is indeed present, even when limited to Shakespeare’s corpus solus.  One particularly notable loss is the separation of previously distinctions between genres that were easily distinguishable in the diagrams.   The Histories still tend to cluster together, and Merry Wives remains an outlier, but most everything else has been moved around.  Interestingly enough, if you look at the three largest branches of the diagram you can note that, for the most part, the Histories are at the top of the diagram, with Tragedies in the middle, and Comedies on the bottom.  Romances are interspersed, with a few other plays also overlapping such a generic genre division.  In a way it makes sense that genre would be preserved if certain punctuation was eliminated, but it is also slightly surprising as well.  In sum, I would say that genre does seem to have an overlying presence in Shakespeare’s corpus that correlates to genre defined by editors, regardless of removing editorial modernization.

Another set of diagrams I included, since I wasn’t sure of what level of Docuscope’s output Prof. Witmore visualized, and because I saw that TimeDate was the only LAT within Narrative Time which meant that I could delete Narrative Time without worrying about losing any more data than what was already in TimeDate.  So below is the comparison picture of Shakespeare’s corpus analyzed with fifty-one Dimensions and then only fifty.  I tried to highlight movement again (by no means a scientific procedure) as I merely tried to look at patterns, several of which were interdependent on each other.   You may notice that there is slightly less movement than with the change in LATs, possibly because we are only deleting one Dimension compared to two LATs.  However, looking at percentages, fifty out of fifty-one is roughly the same as ninety-four out of ninety-six.  Anyways, the deletion reorders the relationships between these elements completely over the course of the two diagrams.  For instance, the first picture has one main branch, with a offshoot of roughly one quarter of the plays.  In the second diagram there is very clearly half and half, which then leads down into four fairly equal quarters.  Apparently modernization’s influence is exactly as Witmore puts it, “what goes in, comes out”.  Catching this aspect of outside influence into the texts will be helpful in later experiments and in creating a solid foundation for digital methods in Shakespearean studies.  Kudos to Mike Witmore and Jonathon Hope for finding this!



Filed under Shakespeare

2 responses to “This Thing of Darkness (Part III)

  1. Pingback: Editors’ Choice: Classifying Shakespearean Drama with Sparse Feature Sets | Digital Humanities Now

  2. Pingback: Classifying Shakespearean Drama with Sparse Feature Sets « Another Word For It

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s