Visualizing Textual Data

Rows of black bars break across an expanse of digital whiteness, each unfolding branch asserting a tenuous yet implicit truth. Graphical data, whether taking form as a mass of scatterplots or a confluence of lines summoned into being, assume the function of pure fact. To some, it can come across as cold, unkind, even patriarchal. At the same time, such data, as fact, betrays the sense in which, as early modern Italian philosopher Giambattista Vico would put it, “the truth is made.” What truths, indeed, do we observe upon observing the “graphs, maps, and trees” that increasingly form the epistemological units of literary and other areas of humanistic study?

This fall, the metaLAB team began to mull the implications of these questions while considering design challenges inherent in presenting computationally-produced data. In the process, we began to realize that whatever information a graph may offer, it generally does not evince insight into its own ontology. To the contrary, data mined using traditional tools and methods arguably conceals as much as it reveals. To this end, we began to wonder: how might we leverage design skills to re-conceive how text mining looks, works, reads, functions?

These questions animate a metaLAB project focused on text mining and multi-authored early modern works. Initiated at Vanderbilt and developed recently at the Folger Institute in D.C. and at Harvard’s metaLAB, Shakespeare, Editor deploys both digital and traditional analytical methods to visualize the “hand” of Shakespeare in a selection of works to which he contributed so as to better grasp his role as collaborator and editor. Primary among these works are Robert Chester’s 1601 multi-author pamphlet, Love’s Martyr, a singular collection containing poems by Shakespeare, Ben Jonson, George Chapman, and John Marston as well as Robert Chester, the traditionally assumed editor of this volume, to whom only this work of print has been attributed.

Shakespeare, Editor?

The hypothesis behind Shakespeare, Editor, which mines and visualizes Love’s Martyr’s authorial boundaries using various stylometric applications, is that portions of text attributed to Chester within Love’s Martyr indicate stylistically significant departures from work signed by him here and in manuscript. As findings suggest, these sections—known as “The Cantos”—align stylistically and technically far more with Shakespeare’s poetry than with the work of either Chester or the other authors who contributed to the volume. As I shared with the lab, one of the most intriguing components of this research was my discovery that sixty years ago, G. Wilson Knight independently reached similar conclusions regarding Shakespearean “doctoring” in this pamphlet using traditional philological methods.

G. Wilson Knight’s The Mutual Flame. “The thoughts and impressions continually suggest the theme of Shakespeare’s Sonnets,” the scholar wrote of Love’s Martyr’s “Cantos” in 1955.

Due to the collaborative nature of the text, as well as the controversial possibility that Shakespeare may have been enlisted as editor or indeed composer of poems not signed by him, the data structures produced by this project beg important questions regarding the epistemological status of the graph. As Matthew put it following my recent presentation, how might these graphs “blackbox” certain knowledges while privileging other, perhaps less telling data? Questions or observations raised by Cris, James, Jeffrey and Jessica likewise drew attention to the nature of data that appears, at first instance, to be self-explanatory, yet upon closer inspection is revealed to be increasingly difficult to interpret and decode.

These questions bear upon Shakespeare, Editor particularly, for our interest is not merely to map author X’s ontological boundaries onto text Y, as revealing as this endeavor may be. Rather, we wonder, how might we develop a lucid methodology for representing conversational, collaborative, and editorial energies within a text as strangely palimpsested as Love’s Martyr? That is, how might we produce data structures that responsibly and accurately delineate such distinctions and progressions while yielding interpretive models that might be interactive, or even attempt to mirror the conversational dynamics of the volume itself?

The only work signed by Shakespeare in Love’s Martyr is “The Phoenix and Turtle”; stylometric and other evidence suggests that he may have collaborated elsewhere in the volume. Image courtesy of the Folger Shakespeare Library.

Further work on imaging Love’s Martyr’s collaborative boundaries will be explored in the coming months. Mainly, work will be done to elucidate the text’s editorial dimensions alongside works Shakespeare was known to have written collaboratively. These efforts build toward metaLAB’s partnership in the Text Mining the Novel Initiative (NovelTM), an international consortium that “seeks to produce the first large-scale cross-cultural study of the novel according to quantitative methods.” Moving forward, metaLAB will seek not only to imagine possibilities for displaying textual data more expressively, but also to explore how the very processes of “mining” works of literature might yield new forms of data altogether, including visual data.

Additionally, we’ll continue to explore the philosophical and practical implications of our research. As I began to consider recently, quantitative research may always yield adamantine impressions, even to digital practitioners at ease with quantitative tools and forms of articulation. However, when treating objects like dendrograms as if they possess ontologies, bound up with yet distinct from the textual bodies to which they have come in contact, we may be better prepared to regard these graphs as something other than crude interpretations of texts to which they refer. Recognizing that these data are also representations of this very content, we may also come to regard these objects with a feeling of heightened curiosity. For in addition to whatever practical knowledge may be gleaned from them, these diagrams present us with a virtual image, a distortion of an event in time that has been instantaneously stamped and destabilized.

Shakespeare, Editor will be featured as a Digital Exhibition at the Shakespeare Association of America’s 2016 Conference in New Orleans, March 23-26.