The Milton Project

This site is a collective effort between John Eckhardt and Katelyn Antolik within the course Slav 1050: Computational Methods in the Humanities at The University of Pittsburgh taught by David Birnbaum and teacher’s assistant Eric Gratta to explore the texts of Paradise Lost, John Miltons 17th century epic.

The central focus of our research is concentrated in the analysis of speech. Our goal is to use computational tools to extract and analyze the speeches from God and Satan. Are there any features of their speech (i.e. diction or syntax) that contrast these two characters?

Topic-modeling has, so far, proven to be a useful outlet for addressing such inquiry. Mallet determines the meaning of a text based on the cooccurrence of words. A user can designate the number of topics that Mallet will create, which are lists of words that are most likely to occur together in the text. We initially took the text in its entirety, and chose to create ten topics. Mallet then iterates over the text, in this case 40,000 times, and determines which words are most likely to cooccur. It then generated ten lists of words, each of which is representative of a particular topic. The results were quite interesting, and you can view them here.