A Digital Commonplace Book
-
Visualizing early modern practices of reading
This interactive viewer aims to provide insight into early modern practices of reading, by placing handwritten notes in an early modern commonplace book alongside the printed book they cite. Commonplace books are notebooks in which readers collect quotes and facts, or 'commonplaces,' encountered in books. This particular notebook, today held at the Princeton University Library, contains a large section dedicated to a popular travel narrative collection written by André Thevet, La cosmographie vniuerselle and published in 1575. By aligning the text shared between them, this site enables the viewer to better understand not only how early modern readers interacted with their printed books, but also how they envisioned the world presented to them in travel texts.
About this project
-
Data collection
This project, which I began as a fellow at the Princeton Center for Digital Humanities in 2023, seeks to visualize the complex relationship between reader and text, as reconstructed by a 16th-century reader’s handwritten notes held at the Princeton University Library, Notes de lecture, ca. 1578–1612 . To obtain the data underlying this visualization, I trained a customized machine learning model to generate machine-readable transcriptions of the handwritten pages. This process consisted of fine-tuning a machine learning model to adapt it for this specific paleographic context, using established tools driven by convolutional neural networks (CNN) such as Kraken/eScriptorium and Transkribus. With this dataset, I then applied programmatic methods for detecting and tracing textual reuse using a very interesting tool called Passim. Along with producing precise data about the locations, content, and context of the citations between the respective volumes, this allows for an easy comparison of the two texts, placing shared passages between handwritten and printed texts side by side in the output JSON file.
One of the problems that I encountered was finding a way to visualize the insights that such commonplace books offer into how early modern readers interacted with printed books. Having a spreadsheet with shared passages, or page XML files that locate passages on the digitized images, doesn't translate this in a way that is immediately accessible.
The prototype for the visualization, which is still a work in progress, uses a simplified version of PAGE XML , the typical output from a handwritten text recognition workflow. I ended up simplifying the structure of the XMLs, eliminating the transcriptions and replacing them with simple line identifiers—'line one,' 'line two,' etc.—which are linked between the two texts.
For a more detailed technical explanation, please see the GitHub README file .