Partners: Writing Rhetoric and American Cultures, Political Science, and the Social Science Data Analytics

Theme: Text Analysis in Humanities and Social Science

Date: 4/9/2015

Time: 3:00-5:00

Location: Main Library, 3 West, REAL Classroom

Increasingly, scholars operating in a wide array of disciplines use computational methods to study digital texts. These digital texts include but are not limited to journal articles, professional proceedings, government documents, novels, websites, and social media (Twitter, Facebook, among others). How can the content of these sources be collected and analyzed to infer the underlying structure and dynamics of human intent or behavior? What computational hurdles and opportunities exist to fruitfully utilize this digitized information in the context of (inter)disciplinary questions?  What leverage does digital text as a medium offer vs. its analog antecedents?  To what extent do computational methods align, complement, or diverge from methods used to study analog text? This LOCUS will gather scholars together to explore these questions in the context of specific research projects and/or pedagogical applications.



Human Rights, Lefts, Ups and Downs: Using State Department Reports to Explore the Evolution of Multi-Dimensional Human Rights Practices and Standards around the Globe

Kevin Greene, Department of Political Science; Michael Colaresi, Department of Political Science

There is an ongoing debate in international relations about the evolution of human rights across the globe. Some recent work concludes that human rights practices have been consistently improving over time, while standards for judging human rights by international organization and researchers have grown monotonically higher. In this project, we attempt to broaden this debate by arguing that the evolution of human rights may not be constrained by a linearly increasing path, along a single dimension running from weak to strong or lax to strict standards and behavior. Instead, we hypothesize that human rights, and thus human rights standards, are multi-dimensional and have specific content that has evolved over time. Thus we expect to observe backsliding on some topics such as torture or rendition while simultaneously improvements in other facets of human rights, such as genocide prevention. To explore this idea we return to the texts of human rights reports that have been used by human coders to create ordinal scores of human rights practices for countries in specific years around the globe. We measure whether the mappings from words to human rights scores have evolved over time and if so, in what ways and for which words. Because of the scale of our problem, we are using tools from machine learning and Bayesian computation to estimate these relationships in a supervised learning framework.

Reading Mansfield Park: Comparative Topic Modeling

Laura McGrath, Department of English; Savannah Smith, Department of English

(Video not available)

The Digital Humanities and Literary Cognition Lab (DHLC) has been analyzing a variety of unusual texts: not only Jane Austen’s Mansfield Park, but also essays written by literature Ph.Ds. after reading Mansfield Park in an fMRI scanner. With our team of undergraduate researchers, we are at the ground-stages of a comparative topic-modeling project, considering the cognitive and linguistic relationships amongst 18 individual subject essays, and between these essays and the chapter of Mansfield Park to which subjects responded. Through analyzing these texts and the relationships between and amongst them, we have already asked a number of fascinating questions about literary reading from a literary perspective.

Topic Modeling Urdu Poetry

Sean Pue, Department of Linguistics and Germanic, Slavic, Asian, and African Languages

(Video not available)

This talk will address some of the challenges and possibilities in topic modeling Urdu poetry, both classical and modern. It will also compare the idea of the ‘topic’ to that of the indigenous concept of the ‘mazmun’, or poetic theme/symbol. The presentation will also consider the possibility of topic modeling as means for comparative analysis of poetry on the acoustic level of sound. The presentation will use a combination of IPython Notebook—a reproducible “notebook” of code in Python—the gensim module—a Python module that alleges to be “topic modeling for humans”—and the Javascript graphics library D3.js for word cloud visualizations.

From a Distance: Affective Responses To Otherness in German Literature between 1779-1961

Anne von Petersdorff-Campen, German Studies

(Video not available)

This work in process seeks to use the text analysis tool Voyant in order to examine patterns in six German literary texts produced between 1779 and 1961. More precisely, I seek to disclose patterns of affective responses to Otherness in these texts. The underlying null hypothesis of this research is that the creation of (in this case) German identity in opposition to a Jewish or African Other has less to do with any one specific quality of the Other, but instead is comprised by the self’s affective response to the perception of Otherness.

PromptMe: Helping Teachers Write Better Assignment Sheets

Laura Gonzales, Writing Rhetoric and American Cultures; Rebecca Zantjer, Writing Rhetoric and American Cultures; Howard Fooksman, Writing Rhetoric and American Cultures

PromptMe is an application intended to help teachers develop better writing assignment sheets. Our system allows teachers to upload their writing assignment sheets to facilitate a discussion about how specific words on thse assignment sheets may be defined. Students provide feedback on instructors’ assignment sheets before writing, hence developing a more contextualized understanding of what they are being asked to accomplish. In this way, PromptMe prompts conversations between teachers and students about writing-related expectations. Currently, we are in the process of testing our system in classrooms at MSU. At the MSU Locus, we will introduce our system, share findings from our research as well as our mockups, and present implications for teaching and learning with technologies.