Really excited to share this new work that appeared in the lates issue of Book History. It is based on our multi-year collaboration with Chad Wellmon (UVA) and Mohamed Cheriet (ETS). It represents our second major output from the “Visibility of Knowledge” project, which aims to use computational methods to study the history of scientific notation.
In our first piece, we reported on the machine learning procedures we used to annotate footnotes in ECCO, a collection of 32 million page images. In this piece we focus on ways in which the field of “document image analysis” can help us see the page as an image: both in terms of the visual qualities of the page, but also in terms of the technical mediations that shape what and how we see the page.
As we write in the piece:
In bringing document image analysis to bear on the history of scientific communication, one of our principal goals is to foreground the “page im- age” as a central unit of historical analysis. Independent of any particular findings that we may uncover over the course of our long-term project, our more immediate aim in this essay is to begin to take seriously the page image as an object of mediation in a double sense: to see the page as an image, that is, to focus on the page as a primarily visual rather than textual object and all of the qualities that attend its graphic identity; and second, to see the page image as an image of a page, that is, as a mediating object of knowledge rather than the thing itself. By combining the insights of book history and critical bibliography with the methodological insights of DIA, we hope to draw scholarly attention to the ways in which what we are seeing (before we begin to read or interpret documents) is first and foremost a representation of an absent artifact. In its most general sense, then, this essay orients us towards the study of the layered mediations of bibliographic knowledge in a digital environment. In doing so, we hope to offer another possible stance for relating to our printed past.
Our hope with this piece is to kick-start more computationally-driven reflections on the material conditions of print history, in the vein of recent work by Paul Fyfe and Qian Ge. So much data-driven work in the humanities (and beyond) starts from a text-centric notion of documents. Page images are seen as an (often corrupted) means to an end.
And yet as a number of recent works by scholars like Bonnie Mak, Ryan Cordell, Christoph Windgätter, or Garrett Stewart have shown, drawing on a rich tradition of bibliographic scholarship, the visual dimensions of books are a crucial facet of their cultural significance.
Our particular focus in this project is tracing the rise and standardization of visual scientific notation practices, like tables, footnotes, illustrations and diagrams. While that will take years to sort out, we thought we would start by offering this framework for researchers to begin to tap into the visual study of documents. We close our piece with some general reflections on the way fields like DIA generate new or at least complementary epiostemological frameworks, which we identify as: “procedural,” “transformational,” “contingent,” and “uncertainty.”