Wilson Special Collections Library, University of North Carolina at Chapel Hill
In March and April of 2020, the library system at The University of North Carolina at Chapel Hill (UNC-CH) in Chapel Hill, North Carolina, was taking steps to prepare for the worsening pandemic and realization that operations for all libraries were about to change drastically for both staff and patrons. Work began in February and March of 2020 to compile ideas for projects that could be adapted to a remote environment with work-flows that could be taught, implemented, and modified online for all library staff who needed at-home work. These projects would be made available to the hundreds of library employees who soon found themselves at home, unable to go into campus to perform their normal work duties. It came as no surprise that many of the projects compiled related to some aspect of updating descriptive metadata or creating descriptive metadata where there was none, for digitized versions of analog materials.
A majority of the projects identified included materials generated from numerous “mass digitization” projects done over the last 20 years in Wilson Special Collections Library. By design, these projects often gave access precedence over item-level description so that more materials could be made available in less time. Most have the most basic and broad “subject, date” format titles and do not include transcriptions of hard-to-read documents. The metadata present with digitized materials was auto-generated from existing XML/EAD metadata created by Archival Special Collections Technical Service staff members.
Many of the materials/collections identified were associated with the Special Collections Library at Wilson Library and this type of work included:
Manual transcription of handwritten materials not currently compatible with available transcription/OCR programs from the Southern Historical Collection
Creation of and quality control for output transcription/OCR files from digitized printed materials
Checking output of auto generated closed-captioning software on digitized video formats from the Southern Historical Collection
Creation of “alternative text descriptions” (alt-text) for digitized images from the North Carolina Collection Photographic Archives
All of the projects/materials identified as possible candidates for work from home activities had benefits for both the library and researchers. Benefits included the ability to improve access for visually impaired researchers (and increase compliance with The Americans with Disabilities Act of 1990) to more digitized materials, as well as providing an opportunity to add to descriptive metadata associated with each image. In my role as the Visual Materials Processing Archivist for all collections in Wilson Library, I was consulted for some suggestions of collections with existing digital content that were minimally described. For several years there has been an ongoing effort in Wilson Library Technical Services, to contribute to a larger discussion aimed at creating more inclusive description and to remediate some of the “legacy description” issues commonly found in finding aids created in the past for collections in Wilson Library Special Collections Library due to outdated descriptive practices. These issues also occur when existing description (sometimes transcribed from printed index or enclosures) is imported and the result is wrapping code around existing data. Often this original description, that was written in the past and likely transcribed from somewhere else, is moved into a new environment for access without being examined very closely.
Working with my colleagues in the (larger) UNC-CH Library Information Technology Services, a very productive project evolved that resulted in alt-text metadata being generated for nearly 10,000 images that are currently online. At times there were as many as 10 people working on this project. They ranged in experience from employees who had been with the University for decades to first year student workers whose initial introduction to archival materials included their work on this project. This project has generated metadata that includes descriptions that will be further evaluated and involved in discussions and policies being discussed and developed as part of a library-wide “Conscious Editing / Ethical Description” forum related to description and content included in materials present in the holdings of numerous libraries on the UNC-CH campus. A majority of the images included in this project come from collections that were largely created by, collected by, and document, white families, white owned businesses, and majority white organizations located in the United States south.
Why does creating “alternative text image descriptions” for images matter?
Some principle reasons why adding alternative text to images is useful:
Allows screen-reading tools to describe images for visually impaired users
This written description of an image appears in place of the image if the image fails to load. (For example, the user, browser, or screen-reader may block images from loading)
Allows search engines to better crawl and rank the website's page
Allows for discussion about descriptive practices, which often involves problematic and offensive language or subjects commonly found in special collections materials
The Alt-Text Work From Home Project created the following types of metadata for images:
Alt-text caption (125 character maximum)
Transcribed text from materials
Format (object/materials scanned)
Front or verso of scanned item
Additional notes
Could the object be considered offensive?
The work-flow crafted included a recorded session in which people involved with the project introduced themselves and the project. Documentation of a “walk-through” was created which included examples of completed work and some discussion of issues that may arise while working on the project. Clear lines of communication were set up, via MS Teams, which allowed for prompt action and communication between participants and coordinators. We held regular “office hours” for several weeks after the project was launched where participants could talk directly with coordinators. It was emphasized that this was an experiment, and that whatever information gathered would be looked at further before going “live” and that all of the work was an improvement over the base-line descriptions currently available to researchers. Nearly all of the more than 10,000 images worked with were arranged and described decades ago with minimal description and access that was often based on subjects (over people) that in some cases may not even be accurate.
Materials documenting the southern United States during the 18th-21st centuries often contain problematic language and inequitable description inherited from legacy practices. One of these practices, that is commonly found with collections containing photographs documenting the American south, like those in Wilson Library, is an assumption that these images, by default, likely depict more white people than people of color. Also included are assumptions made concerning a basic understanding of the history of the American south, in particular the American Civil War (1861-1865) and it’s lasting effects on people of color in the American south. This is also the case when addressing issues including any description that includes race, gender, ethnicity, and economic status. This applies to the donors, creators, or participants included in materials in the collections.
The work from home project team provided guidance and examples to help participants do their work. Work-flows and examples were updated depending on which collection was being worked with, the subjects contained, the era in which they were made, and/or nature of the collection and the creator(s) of the materials. So many of these items were digitized as part of “mass digitization” projects where the goal was to get more materials online, with less emphasis on description. Participants in the Alt-Text Work From Home Project were encouraged to briefly analyze the images and content they generated or transcribed, here is an example of just such an opportunity.
Sample image from the North Carolina Collection Photographic Archives, UNC-CH:
In this instance, an artificially created collection of visual materials documenting North Carolina’s 100 counties (#P0001), includes an image of Mr. Joe Ballard of Salisbury, N.C. This object is from a book containing copies of photographic postcards depicting scenes in Salisbury, NC. Note that the image has some additional information that will be transcribed and incorporated into the alt-text caption. Currently the “alt-text” caption says only the folder title (there are 20 unique images in folder) present in the finding aid: “Folder 1305: Salisbury: Streets, circa 1890s-1950s: Scan 06.”
Sample metadata created from the project:
Similar image from the Rowan County Public Library, Salisbury, NC:
The alt-text project provided an opportunity not only to enhance existing description and include transcription of existing text; participants were encouraged to do a little digging if the subject piqued their interest, or seemed relevant in a way that may have been overlooked in the past. With a bit of online investigation, the Edith Clark History Room at the Rowan County Library in Salisbury, NC provides even more insight into Mr. Ballard’s life as a prominent Black politician at the turn of the century. Also included is a credit for the original image and a circa date of 1908. The participants were provided a space to record this type of information and to include links to information that might help us provide more context. It should also be noted that in 2010 when many of these initial descriptions first appeared online, that the ability to auto-generate even this most basic level of description from the existing XML/EAD finding aid, was a MAJOR advancement and should not be diminished due to existing description being what it was, very bare-bones.
I for one am looking forward to how the metadata created during this project will be used to make these materials more accessible. As the University plans a return to a more “traditional” semester this summer, these work from home projects are all being concluded. The University learned a lot of lessons about serving remote researchers during the lock-down and I know there is much on-going analysis of “what worked” and “what didn’t work.” It is my hope that the important work of generating useful and relevant alt-text for digitized images continues and that the work becomes part of the work-flow for digitizing materials. Having been involved with the initial creation of metadata for many of these images, many of which were digitized over a decade ago, it drives home the point that so much of our work, especially archival descriptive work, is iterative, is always evolving, and we should be ready to re-visit and update when needed.
Patrick Cullom, Visual Materials Section Chair (2020-2021), has been the Visual Materials Processing Archivist, at the Wilson Special Collections Library at the University of North Carolina in Chapel Hill, North Carolina for the last 14 years.