Narrative Atlas is a developing application for computational processing of literary texts through user input. The application compares cultural and personal associations that emerge as different people read text. It generates a system that takes into account how any individual’s unique experiences shape interpretation of text, especially the words that lead to differences in interpretation. The underlying thesis is that understanding of language is both a reflection of our cognitive process and a constant director of our streams of consciousness, our feelings, and our understanding of the world. The project has the potential to reveal specific words and phrases that act as points of distinction between individuals. The project is conducted in collaboration with Leslie Synn, a PhD. candidate at the Graduate Center of the City University of New York. A description of Narrative Atlas was presented at the 2010 Communicating Forms Conference at the University of Chicago.
How it works:
Narrative Atlas is written in python and utilizes the Natural Language Tool Kit to function. The current functionality allows for direct comparison between different texts to determine how the same terms function differently in different contexts. The examples on display compare two literary passages that share a set of visual objects. For example, we have processed two excerpts through Modes I, II, and III—a description of Wordsworth’s experience on Mt. Snowdon in his Prelude and Edgar Allan Poe’s ending to his short story “The Fall of the House of Usher.”
Currently, Narrative Atlas exists as a graphic user interface that allows for base-level Natural Language Processing of a user-selected text within user-defined categories. Our goal is to extend each instance of user categorization of a text into a cumulative process. The program itself is structured to facilitate connections to local or remote databases. This functionality allows choices of a user to be catalogued over the course of the program’s use, providing a dataset that reflects to a degree their unique readership. For example, the categorization of the verbs of a specific text as they relate to the defined actors in will potentially vary from reader to reader. These points of distinction are what Narrative Atlas seeks to aggregate.