Faculty Sponsor: Maryam Gooyabadi and Joseph Slaughter
Jake Klasky
Jake (they/them) is a rising junior (’28) from outside of Seattle, Washington and is majoring in Computer Science and is doing the Data Analysis Certificate. Outside of academics, they play Ultimate on Wesleyan Nietzsch Factor as well as spend time with friends to play analog games. Their favorite is Root: A Game of Woodland Might and Right. When they are back home they love going on short 10-20 mile day hikes with their friends in one of Washington’s three national parks or one of the six national forests, identifying species along the way and enjoying the ambience.
Abstract: Both firearms and Christianity have been enduring actors in the cultural canon of the United States, yet the relationship between the two remains largely understudied. This project applies natural language processing and large language model-based interpretation to examine how firearms-related discourse in American Christian periodicals has evolved over the last two centuries, drawing on a corpus of approximately 99,315 keyword occurrences across 8,444 issues from 11 distinct publications. Each occurrence is classified to filter genuine firearm references from false positives (proper nouns, verbs), then coded for connotation and narrative context (missionary, military, metaphor, cultural decline, advertisement). Using The Atlantic as a secular control spanning the full era, we assess whether shifts in firearms sentiment are distinctly religious or track broader public discourse, analyzing trends in salience, frequency, sentiment, and semantic drift, with dimensionality reduction techniques used to visualize the corpus over time. By combining computational scale with interpretive nuance, this approach offers a more objective and exhaustive method for tracing the changing significance of firearms within Christian communities over time. This project centers on the methodological framework underlying this analysis—detailing corpus construction, keyword classification, sentiment benchmarking, and LLM-assisted contextual interpretation—as a model for applying data-driven techniques to historical and religious studies research.