Visiting fellow at the eHumanities Group in Spring 2014.

Mining Digital Repositories, Trial Record as Historical Source, DH BeNeLux, Surveillance & Society, DH2014, CATCH, the weekly New Trends lectures, and more events that would fill another three lines and my three months with the eHumanities group as a Visiting Fellow this past summer; the frequency of gatherings of academics concerned with computational social science and digital humanities was fantastic, fatiguing, and worthwhile. “Yes,” my department chair said, “you can be out of the country for the last month of the semester,” so long as there was a way to finish out teaching for the semester. While that meant lecturing on narratology and language processing over Skype until 2 a.m. twice a week and arriving many mornings bleary-eyed at Meertens, the intellectual and organizational stimuli (and free coffee) kept me going.


How does one trace borrowings and loanings of sound across culture? How do study at scale the manner in which cultures influence each other? How can event stories be threaded together from across hundreds of news sources? Can museums and other cultural heritage organizations provide ad hoc thematic organizations of their catalogs? What happens if I turn my connectivity off at 18:00 every day? Do n-grams carry syntactic information? For someone who enjoys questions at scale (e.g., one of mine, can narratives of victims and perpetrators of mass atrocities be identified, extracted, and collated from across large-scale collections of witness statements and government reports), working for the summer with the eHumanities group is a good place to be. Add in the density of playgrounds and ice cream in the Netherlands for my then 3-year old daughter (“Dank je wel!” she exclaimed frequently, followed by, “I speak Boston Dutch!”), and it made for a productive and enjoyable summer.


Among the most tangible outcomes was an invitation to participate on a now accepted proposal to the Association for Computational Linguistics (ACL) for The First Workshop on Computing News Storylines [CNewS]; a joint proposal to study manifestations of toxicity and hate speech in online games; and a web documentary, “Digital Humanities in the BeNeLux Region.” The workshop, planned as a one-day event at NAACL/ACL/EMNLP in 2015, emerged from VU’s NewsReader project and Tommaso Caselli, Marieke van Erp, and Piek Vossen, then grew to a broader group concerned with linguistics, narrative, and computation including Anne-Lyse Minard (Fondazione), Bruno Kessler, Mark Finlayson (Florida International University), myself (Georgia State University), Jordi Atserias (Yahoo! Barcelona), and Alexandra Balahur (European Commission’s Joint Research Centre).

The toxicity study is one I have been concerned with for years, as the phenomena is pervasive, pernicious, and undeniably harmful to individuals and society. Most commonly encountered in forms that parallel offline ethnic, gender, ableist, and religious hatespeech, online versions span the continuum from the relatively innocuous angry comment on a YouTube video, to the threats to home and safety accompanying doxing (the revealing of personal information such as home address in a public forum). With Antal van den Bosch (Radboud) and scholars from disciplines as varied as law and game studies, the project looks to bring stakeholders and researchers together for a summit to explore how to study, address, and moderate these manifestations of toxic behavior and hate speech in online games.

The web documentary, enabled by the fantastic drive of Andrea Scharnhorst and Jeannette Haagsma, consists of a set of short, single question interviews with approximately 30 scholars working in Digital Humanities in the BeNeLux region. Introduced by Sally Wyatt and closed by Melissa Terras, the documentary explores definitions for DH, stories of successes and failures, and impressions of the possibilities for societal impact of work in DH. Built as an interactive web documentary, the film relies on the Korsakow System. That tool, itself a novel DH project going back to 1999, places video clips in a network of in and out keywords, rules for connecting those keywords, and interfaces for presenting the content. When well-constructed, as in the work of Florian Thalhofer or Matt Soar, the results are evocative, aggretory, and multi-ideological.

It was a good three months.