By Kiley Jolicoeur
*Image above: Souvenir card from Joe Maloney’s high diving act, undated. Special Collections Research Center, Syracuse University Libraries.
Please note that this post includes many instances of sensitive and harmful language that are discussed in relation to the collection and the remediation work done on it.
The collection – In the summer of 2021, Syracuse University Libraries started migrating our digital archival collections content onto our newly acquired DAM, Quartex. As part of the initial migration effort, a set of collections was identified for complete migration of files and metadata to be available on a new front-end designed for users before July 2022. One of the collections identified as part of this first phase was the Ronald G. Becker Collection of Charles Eisenmann Photographs.
This collection was identified partly because of its large size (slightly over 1400 items) and because of its notable historical and research value. The digital collection is composed mainly of cartes-de-visite and cabinet cards dating from the mid- to late-19th century. Most of the photographs are studio portraits of so-called “freak” performers, people with visually obvious physical differences[1] who performed in specific types of roles with “freak shows” that were associated with dime museums, sideshows, carnivals, and other types of 19th and early 20th century entertainment. Part of the difficulty in describing the materials in this collection is the necessity of including harmful language in description; for instance, “freak show” is a term that denotes a very specific concept and has no proper synonym that can be used in its place.
I began working on the collection in the late summer of 2021 as a graduate student with the Digital Library Program (now the Department of Digital Stewardship), cleaning the legacy metadata and transforming it to fit the new application profile we had recently designed for use with Quartex. There were no plans to remediate any of the metadata at this point, but it quickly became apparent that there were significant problems with not only the quality but also the content of the descriptive metadata, some of which is undoubtedly harmful. I picked this work up again at the beginning of 2022 when I took my current position as the Metadata Strategies Librarian.
The provenance of the collection’s original metadata remains unclear; we’re still trying to identify and investigate any documentation we may have that can elucidate the history behind it. The collection was donated to the Libraries by Ronald G. Becker in 1985 and subsequently digitized by a vendor in 2005, but it’s unclear when the descriptive metadata was created. This uncertainty underscores the importance of documentation not only at a project level, but that also traces the provenance of metadata creation, transformation, and remediation. However, no matter what party was responsible for the act of creation or in what year it took place, it does not diminish the fact that the responsibility for the metadata falls on the Libraries as the stewards of that information. For this reason, the collection was removed from phase one of the DAM migration process so proper attention could be given to its improvement.
The harm – Much of the descriptive metadata makes it apparent that the information professionals working to create the information had taken an approach centered on an idea of objectivity; they created descriptions that interacted only at a surface level with the visual content rather than with the people portrayed or the deliberate composition of the image. Many of the descriptions ignored the names recorded in the transcribed text from the card’s verso and instead used as part of the description “[x] woman” or “[x] man” in place of a known name, leaving that transcription or identification for the end of the description. The first word of this phrase was reserved for a label describing either the person’s physical difference or the name of their role, which were not always the same. For example, in many of the descriptions this read as “midget woman.” This language is obviously harmful for many reasons. Firstly, it erases the identity of the performer and instead presents them as an exhibit without individuality. Secondly, today “midget” is an incredibly pejorative term that should not be used to describe a person. But that being said, the little people who performed in this role specifically preferred to be called “midgets” because it differentiated and defined their performative role, so it is a historically necessary term to preserve.[2] This was also the case for many other roles, such as “giant,” “fat man,” and “bearded lady.” Text on the verso or recto of the cards commonly identified the role of the performer, even if it did not identify them by name, in order to draw the attention of potential purchasers. Each of these individual roles had specific costuming, poses, and photographic composition methods that were used to create the performance; the role of “midget,” for instance, can often be identified by negative space in the top third, top half, or top and bottom thirds of an image, formal clothing, and performers standing with props such as tables, chairs, or taller people in order to emphasize their stature.
My solution to incorporating these terms is to contextualize their usage and to center the name of the performer.[3] In a specific example, an original description read: “Full view. Midget woman standing next to table.” The remediated description takes Rebecca Myers’ name from the transcribed text and reads instead: “Full view of Rebecca Myers, a white woman who performed as a midget, standing with one arm resting on a table with a backdrop of an outdoor scene and gazebo behind her. She later married Reuben Steere and went by Rebecca Steere.” This description centers Rebecca Myers Steere’s name, de-centers whiteness as an assumed default, places the label of “midget” in its context as a performed role, provides visual description of the major elements of the photo’s composition, and provides relevant information as to why her name appears in two separate forms.
This is only a brief look at the work done on the descriptions because its introduction is relevant to the discussion of the titles.
The objects’ titles – The titles of these digital objects matched the titles provided for each item in the collection’s finding aid. Because of the uncertainty regarding the provenance of the metadata, it’s unclear if the titles were created based on the finding aid’s titles, vice versa, or if both were created at the same time. Either way, the titles of the digital objects in the DAMS required attention as part of the remediation process.
Most of the titles reflected the content of the original descriptions. The example discussed earlier was titled, “Rebecca Myers, midget.” This title repeats the same harm of the description, using the role of “midget” without context, where it reads as though it is a descriptor of Myers Steere rather than of the role she performed. In order to align the approach to the titles with the approach to the description, I removed performed roles from the titles and instead disambiguated as much as possible with visual information. This title became “Rebecca Myers standing with table.”
Another complication is that instead of using known names, many of the titles use stage names, some of which are extremely dehumanizing. For example, “Jo Jo the Dog Faced Boy” was the title of a photograph of Fedor Jeftichew, who performed with the stage name “Jo Jo the Dog Faced Boy.” This is an identification problem for the cases of performers who performed multiple roles or characters. But most importantly, it prioritizes the character, or the human exhibit, over the person who was performing the role. For this reason I instead centered the person’s name in the title rather than their stage name and I contextualized that stage name as such in the description field. The title of this particular object thus became, “Fedor Jeftichew, portrait.” The description became, “Full view of Fedor Jeftichew, who performed with the name Jo Jo the Dog Faced Boy, sitting in a chair.”
Titles are, of course, ideally specific and unique, allowing users to distinguish between similar objects. However, I consider our need to prevent persisting (and new) harm to outweigh this consideration, particularly as these are digital objects and have many other access points. With the front-end design we have developed with Quartex’s options, our objects display on the list pages with a title, date of original, format of original, and language. Below is an example from the L. Frank Baum Collection.
Much of this information is not necessarily helpful for disambiguation in the Sideshow Performers Collection. 595 objects include “cabinet photographs” as a format of original value and 450 include “cartes-de-visite.” There are a dozen or so other terms that appear as well, some of which in combination with these two most common terms, but unlike a collection with more varied materials, these terms don’t help to easily distinguish between items. At the moment, roughly half of the collection is “undated,” where a date is not indicated on the original item and has not been determined from context. Even as we address this in the future and try to determine more specific dates for these undated items, it is likely that we will end up with large date ranges spanning multiple decades, and thus that these values will also repeat frequently. Further, most of the items that have text are in English, meaning there is little disambiguation between languages.
So how are users meant to find discrete objects when there is so much repetitive metadata across collection items? We have two approaches to this. The first is that Quartex has built-in full-text searching; even when using the main search bar, the default search includes every indexed metadata field. Thus, every single field that is available to users is immediately searchable without the need to have prior knowledge of search settings or to understand how advanced searching works. With the role term of “midget” removed from the title, if a user wishes to search across the collection for performers who performed this role, they can find every digital object in which the term “midget” is used simply by searching the term because Quartex will identify it from where it’s been included in the description field. This forces a user to confront the ways we think about and describe the roles that these performers were performing, creating an interaction between the user and the metadata. This will also create a divorce between the performer’s identity and their performed role, since the user will interact with the former before the latter when the digital objects are shown in the display. Because this collection is so large, searching is expected to be the main method of interacting with the constituent objects (rather than browsing).
The second approach is serendipity, which I realize may not sound like a deliberate approach since relying on serendipity sounds like it should be an oxymoron. By embracing a situation in which full disambiguation is not possible, we open the door to a situation in which our users will be interacting with objects much as audience members at a freak show would be interacting with the performers, juxtaposing expectations of “defaults” (e.g. whiteness, able-presenting bodies with certain characteristics, etc.) with the reality of both the performed roles and the visual presentation of the performers therein. In her book Staring: Why We Look, Rosemarie Garland-Thomson argues that, “staring becomes a starer’s quest to know and a staree’s opportunity to be known. Whatever or whomever embodies the unpredictable, strange, or disordered prompts stares and demands putting order to apparent disarray, taming the world with our eyes. Because we are all starers, knowledge gathering is the most productive aspect of staring in that it can offer an opportunity to recognize one another in new ways.”[4] It is also my hope that viewing objects that have only partially disambiguated titles will lead users to confront the act of voyeurism that is inherent in viewing materials of this kind, since they are facing this juxtaposition of expectation and reality.
The finding aid – The conversation around this remediation of course grew to include the finding aid and was addressed at a meeting of the Special Collections Research Center’s Social Justice Team. Because the Department of Digital Stewardship will be moving to ARK IDs to provide stable URLs in the future, linking each titled item in the finding aid with its digital surrogate is an eventually attainable goal but not currently feasible. The identifier on the items in the finding aid will also be in the digital object metadata, which allows users to match the archival object with its digital surrogate, but this is not an ideal way to present the information since it demands more knowledge and work for the user. Instead, once the newly remediated metadata has been reviewed and is ready for publication, the new titles will be added to the finding aid with accompanying descriptions to further disambiguate them.
The collection title – Lastly, the title of the digital collection itself also received attention during this process. The archival collection’s name, the Ronald G. Becker Collection of Charles Eisenmann Photographs, had initially been the digital collection’s name as well when it was made available on our legacy digital collections site. Part of the work I did as a graduate student was to isolate the photographer’s information based on the photographer’s marks on the cards in order to separate that information into its own fields and improve searching. What I found was that Charles Eisenmann and his successor Frank Wendt were only responsible for creating about a third of the photographs in the collection. The rest had been produced by many other photographers in the United States and Western Europe, some of whom were famous and some of whom operated only locally and photographed the performers as they traveled. This caused confusion for users, as they would expect to see a collection of only Charles Eisenmann’s photographs.
Because of this, as well as the fact that the performers themselves are the subjects of all of the photographs, we decided to rename the digital collection the Sideshow Performers Collection.
Making work visible – Particularly when renaming a digital collection with a title that differs from its archival counterpart(s), understanding and navigating the relationship between archival and digital collections can be confusing for users and staff. In our digital collections, we don’t view the digital collections themselves as digital surrogates for archival collections. Since many of the digital collections are curated from their archival counterparts but do not represent the full archival collection, representing them with DACS-based titles such as “L. Frank Baum Papers” is confusing for users when the digitized materials don’t reflect what they would expect from that name, such as in this case where the digital collection is almost entirely composed of photographs and other images. For this reason, we title our digital collections without material information (in this example, “L. Frank Baum Collection”) and provide content information in the collection description that displays when a user views the digital collection. Some of the digital collections are also combinations of digitized material curated from multiple different archival collections to provide a better user experience, so choosing a title that does not suggest a certain kind of material and encapsulates what is in the digital collection is especially important and sometimes especially challenging.
The finding aid link for the corresponding archival collection(s) is always present in the description of the digital collections and in the metadata of every object associated with that archival collection. The title of the corresponding archival collection is also always present in the metadata, included in the Source Collection field that also displays in the list view, before the object is even selected by the user for full viewing.[5]
Because the Ronald G. Becker Collection of Charles Eisenmann Photographs is such a popular collection, there has been some confusion about the relationship between the archival and digital collections. The Sideshow Performers Collection has a landing page users will see before viewing the collection’s materials, in which the relationship between the two collections is defined.
This landing page will be updated with more information about the remediation project once it is complete and the remediated metadata and the objects have been published to the site, removing the redirect to the legacy content.
Plans have not yet been made for what information to include in the finding aid, but the relationship between the archival and digital collections will be noted, particularly since a link to the digital collection will be included in the finding aid.
Documentation about the remediation process, including discussion of the decisions made, will be maintained in our departmental workspace.[6] Because of the space limitations of the landing page, where only a limited amount of text can be included before the “Search the Collection” button is below the scroll cut-off and thus less obvious to users, we are brainstorming ideas of what information we want to share and how we can best make it discoverable to our users. Ensuring that our work is as visible as possible is of great importance to me, as it invites feedback that can inspire improvements and may prove useful to users, researchers, and other professionals. I also believe that metadata work should be discussed loudly and publicly so users learn more about the processes behind the scenes and the creation of information becomes better understood as a human product created by professionals with the requisite knowledge, and that it will always be limited and inherently biased because it is a human product.
Further work – The remediated metadata will be evaluated by members of the Special Collections Research Center and feedback will be incorporated before it is made publicly available. I look forward as well to receiving feedback from students, faculty, and researchers as they interact with the new information. I am sure there will be need for improvement in the future, and I welcome that opportunity.
Metadata work should never be considered “complete.” One of my more general next steps is to create a remediation plan and workflow for continued assessment and reassessment of the digital collections we steward that I’ll present to members of the Department of Digital Stewardship and the Special Collections Research Center for workshopping. The Special Collections Research Center’s Social Justice Team is also currently working on developing guidelines for conscious description for use in creating new descriptive material and remediating existing descriptive material.
[1] Please note that I use “physical differences” rather than “disabilities” deliberately. Not only has the cultural concept of disability changed over time and is too complex of a label to ascribe to the social perception and self-identification of these performers, but the differences also encompass visual traits such as skin color.
[2] For a more in-depth discussion on this, see Robert Bogdan’s Freak Show: Presenting Human Oddities for Amusement and Profit, 1988.
[3] If you’re interested in a more thorough discussion of why I argue for this solution, I discuss how an approach grounded in care ethics can and should be used to create relationships between memory workers and people represented in archives in my DCMI 2022 paper, “Practicing Care: A Look at the Application of Care Ethics to Metadata Creation and Remediation.” View here.
[4] Garland-Thomson, Rosemarie. Staring: How We Look. Oxford: Oxford University Press, 2009, 15.
[5] The list view can be seen here: https://digitalcollections.syr.edu/l.-frank-baum-collection. A full object view can be seen here: https://digitalcollections.syr.edu/Documents/Detail/l.-frank-baum-with-principal-actors/22198.
[6] The legacy metadata will also be maintained and made available to researchers upon request.
Kiley Jolicoeur is the Metadata Strategies Librarian at Syracuse University Libraries.