This post is a continuation of the previous post’s project, although the data put into the diagrams is slightly different. Now the data selected is only the ‘w-o.txt’ plays, the plays without their main character’s lines, and the ‘.txt’ files of the character’s words. In addition to the departure from the last post, these diagrams have been labeled according to genre following this format: blue ‘^’ for Comedy, red solid square for the main character’s lines, orange ‘+’ for History, black triangle for Tragedy, and light read empty square for Late Plays. The methodology for creating the diagrams is otherwise the same as noted earlier.
When the data set is run, the use of Cluster-level analysis, on the left half, portrays a fairly robust set of groupings consistent with genre divisions of the First Folio with a main group of Tragedy and Comedy and a slightly spread out group of History. The Late Plays do not seem to exist as an entity in this view and the characters do not really form major groups either. But in the LAT-level analysis, much of this seems to be turned on its head. For instance, the character’s lines form two large groups and the Comedies and Histories tighten ranks as well. However the Tragedies all but disintegrate into the background which is a significant departure from anything that we have seen previously. But why would the separation of characters’ lines from plays create this kind of effect?
One reason could be a mistaken understanding of genre as a modern audience. In 1714, Nathan Rowe said that Shakepeare’s
plays are properly to be diftinguifh’d only into Comedies and Tragedies. Thofe which are called Hiftories, and even fome of his Comedies, are really Tragedies, with a run or mixture of Comedy amongft ’em. That way of Trage-Comedy was the common Miftake of that Age, and is indeed become fo agreeable to the Englifh Taft, that tho’ the feverer Critiques among us cannot bear it, yet the generality of our Audiences feem to be better pleas’d with it than with an exact Tragedy. The Merry Wives of Windfor, The Comedy of Errors and The Taming of the Shrew, are all pure Comedy; the reft, however they are call’d, have fomething of both Kinds. (Rowe XVII)
The evidence in Rowe’s favor is present in both diagrams. In the Clusters-level, Comedy of Errors and Taming of the Shrew are paired and all three join in the LATs-level diagram. In addition the groups of History intermixed with Tragedy can be explained by this in the LATs diagram as well as the small group containing Midsummer Night’s Dream at the bottom. However Rowe’s theory is complicated by the presence of characters’ lines as distinct from the plays themselves since it begs the question of what exactly makes a text either Comic or Tragic since I would assume that genre relies on the main character, but does it?
If we return to the hypothesis of the previous post, size could also be a reason for associations in this diagram, even if it was not for the other diagram. In these dendrograms, Tragedy could be dissolved through the simple range of words that the genre encompasses. If you look at the chart below, Tragedy has a range of over 13,000 words between its longest and shortest plays compared to 8,000 in History and Comedy, and 9,000 for the Late Plays. The discrepancies on consistent size could be coupling with any kind of cross-genre similarities to pull apart Tragedy from itself. For example, Othello has long been noted to be near Comedies in many experiments, yet it ends in death and is listed as a Tragedy in the First Folio. Other examples of this effect could exist in more subtle ways and or enlarged by separating a main character that was particularly tragic from the play as a whole. It would also explain why a large number of characters’ lines group together.
Overall, I am hesitant to immediately assume a drastically different interpretation of Shakespeare’s genres than those in the First Folio which I have based many decisions off of, but at the same time I am still doubtful of the significance of length in relation to understanding the difference between these plays. There does seem to be validity in both approaches yet neither seems able to appropriately address the instance and evidence that is in front of us. It would seem that we have information in these dendrograms that is representative of many different factors but, if that is true, how are we going to discriminate between one influence from another?