New Pedagogies for Quantitative Text Analysis

The Text Analysis Group focused on pedagogical applications of text analysis using statistical methods over the course of summer 2017. The team is comprised of faculty from various departments at both Grinnell College and University of Iowa who are interested in exploring pedagogical applications of text analysis tools. The collaborators included Erik Simpson (English) and Pamela Fellers (Statistics) from Grinnell and Frederick Boehmke (Political Science) and Paul Dilley (Classics) from Iowa, with project management and participation from Matthew Hannah (Digital Bridges Postdoctoral Fellow). They met over the course of the summer to plan how best to implement these tools into their individual courses.

photo from British museum of the history of computing in Bletchley ParkErik Simpson spent the summer working with Christina Brewer (Grinnell undergraduate) to develop a course on digital studies in literature (ENG 295: “Lighting the Page: Digital Methods in Literary Studies”) for spring 2018. This effort is the first step in a project supported by Grinnell’s Innovation Fund and co-proposed by Simpson and Sarah Purcell, who is developing a parallel course in history. For Simpson, the collaboration this summer involved 1) building a frame of information and information theory to introduce the course and create a reference point for the whole semester, 2) evaluating a variety of approaches to text analysis and topic modeling, and 3) connecting a series of studies and theoretical readings to the practical exercises that Brewer and Simpson have been working on in Grinnell.

Pam Fellers worked with two Grinnell College Students, Maggie Remus and A.T. Tambay, on the process of text analysis. The students’ projects focused on the statistical components of quantitatively analyzing text data in a particular application. For Fellers, the collaboration involved identifying sources and tools that others used or suggested, as well as gaining insights into the work of digital humanists and the questions they ask. This information helped her better think about how statistical components fit into the larger picture of digital scholarship. Through conversations with the Text Analysis Group, Pam imagined ways that statisticians can contribute to the conversation and how to communicate to those without as much statistical background.

Fred Boehmke and Paul Dilley are team-teaching a fall Big Ideas course on “Information, Society, and Culture,” and they used the Text Analysis Group to explore academic and popular literature related to text analysis (e.g., Nabokov’s Favorite Word is Mauve) as well as appropriate tools to allow students in introductory (Voyant) or advanced (Python) courses to analyze and model text as data. They discussed ways to get students engaged with the text, including the possibility of using established authors, social media, popular contemporary books, or students’ own writing samples. We also talked about progress in developing our various courses. Dilley used the collaboration to begin developing a module to introduce students to the basic concepts of information, data, and statistics, which now contains numerous student exercises using dice to learn about frequency distributions, measures of central tendency and variability. These meetings have been especially helpful for Boehmke as a social scientist to learn more about techniques and software tools to teach digital humanities to students. Dilley’s contribution focused on topic modeling through the recently released R application ToPan, which has a relatively simple interface that can be used in the undergraduate classroom. Dilley also plans to create a class twitterbot, using Cheap Bots Done Quick, to follow tweets which contain the phrase “data matters,” one of the major “Big Ideas” for the course. Finally, Dilley and Boehmke also discussed potential in-class exercises for a module on recommendation algorithms.

Frederick Boehmke, Paul Dilley, Pam Fellers, and Erik Simpson were supported in this work by a Digital Bridges Summer Collaborative Grant. Boehmke is also involved in a Digital Bridges collaboration with Timothy Dobe (Religious Studies, Grinnell) on Building Movements for Social and Religious Change. Paul Dilley was a presenter at the 2016 Digital Bridges Summer Institute and is one of the leads on the Digital Bridges-sponsored Mining the Renaissance project. Erik Simpson is the Grinnell-side Principal Investigator of the Digital Bridges grant.