According to our reading for this week, text analysis has its roots in literary history and criticism. Whereas close reading involves the careful study of a single text, distant reading refers to the large-scale analysis of many texts or collected works. The goal of distant reading is to uncover overarching trends, patterns, and concepts that would not be visible through close reading alone. Crucially, this method relies on computer technology.
What was particularly fascinating in Buurma and Heffernan’s article, “Search and Replace: Josephine Miles and the Origins of Distant Reading,” is the revelation that distant reading actually predates the digital era. The article highlights the work of Josephine Miles, an English professor at Berkeley during the 1950s, who used machine methods to produce a concordance. Her work came before Roberto Busa, often credited as the father of computing in the humanities for his index of the texts of Thomas Aquinas. Miles’s work and methods show how scholars were already imagining ways to analyze literature on a larger scale before computers became widely available.
Our readings also connected these early debates about text analysis to contemporary conversations about Generative AI. Mark Riedl’s “A Very Gentle Introduction to Large Language Models Without the Hype” explains how large language models like ChatGPT actually work by predicting text. Because these models are trained on large amounts of online material, they inevitably absorb the biases, stereotypes, and misinformation present online. Their predictive nature also means that accuracy is never guaranteed. Another piece, Cate Denial’s Substack post “Why I’m Saying No to Generative AI,” examines the ethical implications of AI in the classroom. Denial and her students concluded that generative AI has little value in historical work. Learning about the progression from early concordances and distant reading to modern AI systems was fascinating. The readings illustrated both the promise and the complications that come with applying computational methods to language.
In my own brief experimentation with Voyant, a web-based tool for distant reading and textual visualization, I analyzed Jane Austen’s collected works. Voyant quickly generated word-frequency counts, topic models, and other visualizations that helped show larger patterns across the texts. I found the tool intuitive to use, especially since it requires no coding. Features like the Word Cloud and Dreamscape were particularly helpful in identifying trends and relationships. I think digital tools like Voyant offer support for qualitative research in the humanities by showing large-scale textual patterns and relationships over time and across a collection of works.