Like I mentioned previously, I began this project tired, under a time restraint, and with no idea of where it would lead me. However, I had hope. At a few research meetings previously, Bill Blake had mentioned doing some work with Docuscope in the sense of Hamlet without the Prince. While I had never really seen the result of his work, the idea of a play without its main character or main characters enticed me. It didn’t take me long to find my favorite play, Romeo and Juliet, and sit down at my computer. Referencing my second edition of the Riverside Shakespeare, I cut and pasted lines spoken by each of the characters throughout the play until I had, technically, two different plays. I retained the initial text file, split it up by Act again, and then created a new text file for the whole play, without any of Romeo or Juliet’s lines. I also separated the Acts again, without their lines, and created two new files; one with all of Romeo’s lines and with Juliet’s. The images I came up with are below. The methods remained the same from the previous post: Hierarchical Cluster, Best Guess Analysis, Ward’s test, Colored Coded Dendrogram with a Distance Scale performed at a Cluster level analysis of Docuscope’s output.
All of the pictures are marked with asterisks where attention should be directed; in this case, Romeo and Juliet’s lines as well as the whole play with and without the two title characters. The first picture was an, what I felt to be exhaustive, diagram of various relationships in the play. All of Romeo’s and Juliet’s lines are labeled plainly as Romeo or Juliet and all of the act and scene divisions in which they speak have a label such as Romeo Act2.2 (Act Two, Scene Two). The whole play is again followed by rev and all of the play’s lines, besides the two characters, is followed by w-o. This diagram shows the multitude of different relationships all of these pieces can have with another, however it should be noted that both the whole play and the whole play with lines missing cluster closely together and both Romeo and Juliet are outliers of their respective branches (i.e. they don’t cluster closely to just one other data point). Just as well, the Act divisions, with and without certain lines, all cluster closely together. The result of this data was both encouraging and disconcerting as it provoked the question of just how important the lines of a character or multiple characters are in a play. (Even when the play is named after them)
Instead of going further in-depth though, I decided to work with what I already had. The second diagram is Shakespeare’s canon with four significant points of interest added: Romeo’s lines, Juliet’s lines, Romeo and Juliet, and Romeo and Juliet w-o Romeo and Juliet. Looking at the placement of the asterisks, it can be seen that Romeo and Juliet and Romeo and Juliet w-o still cluster together, Romeo remains on the outskirts of that same cluster , but Juliet is moved to a different cluster. This result is interesting as it appears that even the rest of Shakespeare’s corpus cannot separate Romeo and Juliet with and without.
After this result, I decided to do what Prof. Witmore later named the “kitchen sink” method. In this, I decided to throw all of Shakespeare’s texts together again, with Romeo and Juliet’s lines, but also with Middleton’s, Dekker’s, and Jonson’s respective corpii. While the minuscule writing detracts from too close of an analysis, the asterisks point out Juliet, Romeo and Juliet and Romeo and Juliet w-o, and Romeo in descending order. Again, both of the R&J (Romeo & Juliet) plays don’t separate however multiple Dekker plays now divide Romeo and Juliet from the R&J plays. Juliet is again an outlier on the cluster and doesn’t group with any particular point, but Romeo has moved to a different cluster and is actually grouping with Dekker’s play Old Fortunas.
Perceiving that Dekker’s plays were the ones that mattered when looking at these four particular data points, I decided to remove Jonson and Middleton’s works. This final image provides a closer view of what I was interested in, and keeps the same features concerning the four points of interest that I mentioned above. However, it also reveals significantly fewer Dekker plays separating Romeo from the R&J plays. With this, I concluded my portion of the conference and my major work with Docuscope for a while. It was inconclusive, but the aim of my part of the presentation was both to highlight the possibilities of this kind of work at the surface level and segue into Prof. Witmore’s conclusion (it was also to get it done before the conference itself). In the end, it worked out.