There has been quite a bit of conversation of late about textual analysis and topic modeling in literary studies (see here, here, here, and here for a handful of examples). The availability of tools like Voyant and Mallet have made it possible for digital humanists to begin work in textual analysis and topic modeling quickly and with little/no institutional support. These tools also allow us to give a fresh dh twist to traditional ways of practicing literary studies. Natalie Houston put this nicely in her talk at MLA last year:

Our method, quite simply, as literary scholars, is to pay attention to patterns. Digital tools offer us computational power for conducting analysis far beyond our human limitations. Such tools can offer us new ways of understanding the material places of Victorian poetry through analyzing patterns in the metadata, page images, and linguistic layers of the digitized text.

In light of all this, I’d like to propose a session in which we discuss the value of textual analysis and topic modeling for digital literary studies. The session could go in a number of different directions, including:

• an exploration of a specific tool like Voyant,
• a conversation about use-cases and current projects,
• a more meta conversation about how and why we might use these tools, and the questions they invite and/or foreclose. I’m particularly interested in talking about the recent meta-analytical work Andrew Goldstone and Ted Underwood have done with PMLA and the new avenues it might open up for understanding the work we participate in, the culture of a particular journal, and the shape of a given field: What can topic models of PMLA teach us about the history of literary scholarship?

There is a gulf between the needs of industry writers and the needs of researchers and academics. One requires very orderly processes of feedback and comments, while the other is a messy connectivism melee of ideas and attribution.

I still recall a drawing of a 16th century scholar (though cannot find it) who has several books held up by mechanisms, cross referencing and comparing the texts. I’m reminded of my own multi-monitor display, and wonder how augmented reality and future devices will change even that modal paradigm.

The closest in terms of present-day systems to this aspiration of connectivity in knowledge remains Wikipedia, though the links (clicking text which changes everything) often hide the connections as much as they provide ways between ideas. Ted Nelson’s Project Xanadu set out to realize the connections in a spatial way, though the system as envisioned was never completed. Giorgio Guzzetta’s proposal for this year’s THATCamp MLA talked of a scholarly operating system, citing the various tools and programs that would allow tweaking and cajoling Linux tools and systems into the specialized needs of academia, and more specifically the humanities. These tools are constantly moving, their combination volatile and their practicality at least uncertain. The primary technologies I currently use are Microsoft Word, Mendeley Desktop manager (which does a great job of helping manage citations and PDFs/Notes), but I (and most others) still feel there should be something more.

I’d like to be a part of a session that addresses these deeply embedded dissatisfaction with the ways that technology supports the fundamental process of writing and scholarship. Who is best suited to standardizing these solutions, who has the time and capabilities, and further, since we are professionals, what would we be willing to do to show there is a market?

Please weigh in on the comments, and I’d be happy to combine talk of semantic web, real-time collaboration tools with the existing applications and workflows.