Tess Dejaeghere

As part of my three-month internship at the DH Commons in 2020, I worked on both the DH Commons and the Digital Humanities webpages in close collaboration with Laura Ulens (my fellow intern) and Merisa Martinez, our supervisor and the founder of the Commons. We started from scratch, with a limited team, no physical space, no budget and no reputation to fall back on. Additionally, the Covid-19 pandemic threw us a curve ball, and led to us having to to coordinate all our activities using Skype ..

The paper presents research on activism in metal lyrics by making use of computational methods. Based on literature, it is assumed that activist markers will be shrouded in gory rhetoric, and that ecological and social activism will show an increase in the past decade. Terminological markers of ecological, political and environmental activism were extracted from a dataset scraped from the Metal Archives website. Using TF-IDF, these were depicted as word clouds and network plots. A topic model scatterplot was made using the Empath Scattertext tool. Ecological activism was surprisingly not characterized by gory rhetoric, as opposed to political and social activism. In a next step, a rule-based activism classifier was built based on these findings and the ranked keywords of a set of topic-specific Wikipedia pages. The activism classifier eventually attained 67% accuracy on a manually collected test set. The results of the classifier on a comprehensive Kaggle Metal lyrics dataset were depicted as a time series, and a continuously increasing trend in ecological activism was noted, partly confirming our hypothesis. While the classifier in its current state is not ready to be released in the wild, this research represents an interesting first stepping stone towards more fine-grained analysis of activist narratives in heavy metal lyrics.