Performing Archive

Distant View of the Image Collection

David J. Kim, UCLA

This network visualization began as an attempt to provide a different view of Curtis’ documentation of the subjects whom he fastidiously organized by their tribal affiliations. One approach was selecting a few notable images as evidence fit for “close reading” of their individual details that may attest to a set of conventions that Curtis’ used in his construction of the Native American identity in the early twentieth century. The result of this approach is Ken Gozales-Day's series of exhibits, as well as the video edited by Heather Blackmore. The network view of the images came out of our curiosity with what “distant reading” of the collection has to offer for our analysis.5 While both potential in-class materials address a similar set of critical questions regarding the images, they provide different pathways of considering them in practice.

We began with a few broad questions that served as our critical/theoretical aspirations for our visualizations: What “types” of subjects did Curtis’ see and privilege to document his subjects? What "data" and "information" does Curtis collect and communicate through the images? Since Curtis himself only provides titles for the images but rarely specific descriptions of the subjects, is there a way to more closely approximate what one sees in the images instead of taking his titles as given referents? Following Willard McCarty's concept of "knowledge modeling," defined as "a continual process of coming to know by manipulating representations," can the heuristic process of describing and categorizing the images for the purposes of locating any patterns lead to a different understanding of the materials?6 Is there any teaching and learning value in the process of manually creating and curating this set of comparatively "small data" as well as "big data"?7

We considered "network analysis" as a way of providing a different framework for Curtis' archive, as it has become one of the most productive data-driven approaches in the digital humanities for visualizing known connections and for supporting new discoveries.  By leveraging the increasing availability and the computability of large datasets, many projects have developed sophisticated algorithms and visualization modes to offer many new insights into various topics ranging from the "Republic of Letters" in the period of Enlightenment to the contemporary Research Oriented Social Environment (RoSE) in the rhetorical form of "nodes" and "edges."8 However, as Elijah Meeks has recently noted, for the humanities-oriented research, “network representation” is often the more accurate description of the approach taken, especially for those projects that do not aspire to the high-level quantitative analysis of "big data" usually associated with network analysis.7  By "representing" The North American Indian as a network of tribes, Curtis' images, and the subjects represented in those images, this experiment is, in a sense, a simplified visual "model" that reflects our critical priority on unveiling the constructedness of Curtis' archival or documentary mode of signifying the "vanishing race" in the early twentieth century.  

With our set of theoretical questions and Meeks' helpful reorientation of how the network is visualized for our purposes, we developed the following goals for this experiment:

1) seeing the collection of images from a distance; 2) visualizing in some graphical form our thoughts behind the need to separate the Curtis’ given titles of the images (ex. “Typical Navaho”) and what the images contain (“[Old Man in Native Dress]”). We relied on the Library of Congress’ description of each of the images as much as possible, and when it wasn’t available we supplied our own approximation of the physical attributes of the subjects, focusing on gender, age, clothing and the activity the subjects are performing (ex. "Storytelling") when discernible. 3) noticing any patterns in his selection of the subjects he chose to represent each of the tribes; 4) supplying searchable categories for the images for the volumes we have studied.

["Typical" Images in The North American Indian]






5. For a recent discussion on "distant reading" and helpful introduction to Franco Moretti, see the introduction and chapter 5 of Stephen Ramsay, Reading Machines: Towards Algorithmic Criticism (Urbana, IL: University of Illinois Press, 2012); as an approach for analyzing a large set of image data, see Lev Manovich, Jeremy Douglass and Tara Zepel, "How to Compare One Million Images" (2011).↩
6. The concept of "small data" as a different set of practices than "big data" was first brought to my attention by my colleague Amelia Acker. See Context and Collection: A Small Data Research Agenda.↩
7. Willard McCarty, “Modeling: A Study in Words and Meanings,” in A Companion to Digital Humanities, eds. Susan Schreibman, Ray Siemens, John Unsworth (Oxford: Blackwell, 2004).↩
8. For an interdisciplinary discussion of the process of developing RoSE, see Chuk, Erik, Rama Hoetzlein, David Kim and Julia Panko. “Creating Socially Networked Knowledge through Interdisciplinary Collaboration.” Arts and Humanities in Higher Education, 11. 1-2 (2012).↩
9. https://dhs.stanford.edu/algorithmic-literacy/learning-network-analysis-and-representation-with-a-pedagogical-toy/↩

This page has paths:

This page references: