Exclusive to M/m Print Plus

Finding the Modernist Archive: Why UX Matters

Using the modernist archive requires finding it first. The modernist archive does not live in one collection at one repository, such as a single university special collections department or one pivotal private library. Rather, the modernist archive is a term used to conceptualize a networked set of collections across many repositories in the United States or abroad.[1] The fact that the modernist archive is dispersed rather than centralized is critical because each institution’s holdings are more or less discoverable based on local application of user experience (UX) principles. Weave, a Journal of Library User Experience defines UX as employing a variety of methodologies to inform improvements to physical and digital space so that the user can easily access collections and services. Th US Department of Health and Human Services provides an overview of UX basics that, adapted for libraries and archives, would require repositories to identify their users, what they want, what skills they have, and which they don’t. According to Coral Sheldon-Hess, when UX is properly implemented, users of all levels of expertise can more easily access what they need. When UX is ignored or poorly applied, users are more likely to perpetuate pre-existing archival silences as well as less likely to have successful searches.

Poorly designed finding aids inhibit the use of archival content. For example, Suzanne Woods has argued that women writers, since they were outside the canon, have had to keep being discovered; as women’s papers were not well-described, each new generation of scholars needed to relocate them. Restrictive canons are reinforced when poor description adds insult to injury.

Although this blog post identifies two interventions I argue would lead to substantial improvements in the UX of modernist archives, ultimately repositories must continually seek feedback on their patrons’ experiences to enhance their current interfaces rather than simply following these or other recommendations. Collaboration is critical. Repositories must better understand their researchers’ needs to design better interfaces while researchers should improve their understanding of the technical and professional challenges faced by those who work in archival institutions. 

Add Accession Records to Archival Discovery Systems to Highlight Archival Silences

If you are acquainted with archival theory, you will know that what is in an archive must be discussed in relation to what is not present. Kate Theimer has summarized how scholars have theorized these absences as archival silences. Archival silences occur when content is never created, destroyed, not provided to a cultural heritage institution, not described/stuck in a backlog, not described adequately, and/or not promoted sufficiently to its creators and/or intended audience.[2] Failure at any step will keep a collection from gaining the prominence it needs to be used to shape research and discussion, but UX principles can help solve silences.[3]

The ideal archival discovery system makes visible what it represents and omits. As a researcher, if I am searching for the location of a modernist author, I want to know where her papers are held and if there are any papers on her work or life that have been acquired, but are not yet accessible. Ideally, I also would want to know when papers that I cannot reach will become available and the reason for the wait, such as if the papers are in arrangement and description, undergoing digitization, or under restriction. Such information is rarely provided, however, because this information is usually kept as internal documents within each archive. Private owners of archives are likely to contribute metadata for their materials to a national discovery system. Recently acquired collections may merit a press release, but not all repositories create press releases for their incoming materials. Even if press releases are common at a particular archival institution, such publicity is confined to collections that would net significant public interest; not all modernist archives do. Repository websites may note if materials are not available and why, but I have observed that this information is more likely to be provided if accessibility is limited due to structural constraints, such as a library’s renovation or a collection’s temporary location in offsite storage, rather than due to internal workflows or donor preference.

Nevertheless, adding accession records to discovery system whenever possible would add more nuance to archival discovery outcomes. Too many researchers think only about what is findable, moving directly from the question “Where are collections related to X” to “What is in X collection” while bypassing the necessary query “Why are only these types of collections all that I am finding when I search for X?” Jason A. Clark has observed the importance of integrating anticipatory design into UX. Providing minimal descriptions of acquired collections to archival discovery system prior to the completion of full finding aids would help meet needs researchers do not even know they have.

These short descriptions could be developed from the content of accession records. Accession records are files that can include information regarding the circumstances of acquisition such as the deed of gift or purchase agreement as well as information on potential restrictions. Although currently repositories use accession records for their own record-keeping, making this information public would substantially improve the UX of modernist archives by better allowing researchers to identify archival silences. 

Create a National Archival Discovery System to Improve Cross-Institutional Discovery

While originally institutions kept their finding aids on paper, now many repositories place their finding aids online through Encoded Archival Description (EAD). EAD is a broadly-adopted, community-based standard for structuring data invented by Daniel Pitti at the University of California, Berkeley in 1996 and implemented as EAD1 in 1998. In 1997, the archival field prepared for EAD implementation by dedicating the Summer and Fall issues of the American Archivist to the topic. Since then, EAD has adapted to changing technology and evolving information needs while aiding researchers’ discovery by allowing each section in a finding aid to become a discrete unit of metadata. This shift has allowed researchers to find aids with similar descriptions at the same institution or at other repositories, facilitating unexpected research outcomes.[4]

EAD seems to involve a simple transformation from a Word document to one that is encoded for digital discovery, but the process is complicated. To describe this process simply, creating an EAD finding aid first requires reviewing the original aid, if one exists, to see how well these initial descriptions were made. Often, finding aids created before EAD standards were implemented contain idiosyncratic features that do not fit established metadata standards. Next, EAD finding aids must be transformed into a properly encoded Extensible Markup Language (XML) document. XML uses a similar tagging structure to HTML, but is not as rigid. HTML is controlled by browsers that dictate how tags are expressed, while XML is more fluid; its code defines structural relationships for discovery rather than how the end product will look within a browser. Furthermore, XML allows coders to create their own tags, a significant benefit as scholars generate new topics for tagging. Third, creating an EAD finding aid is not the same thing as publishing it. Without publication, no one can find the aid. To publish an aid, an archivist can use a variety of methods, from transforming the EAD record into an HTML document to loading it into a management system that does the work for them.[5]

Once an EAD finding aid is published, ideally researchers could find it through a single discovery system. The problem is that this system does not exist. As a result, researchers who want to search all American repositories will have to patch together information found through ArchiveGridArchives West, Arizona Archives Online, Online Archive of California (OAC), Rocky Mountain Online Archive, Texas Archival Resources Online, and Virginia Heritage as well as individual institutions’ websites.[6] Each of these discovery systems has its strengths and weaknesses; a comprehensive evaluation of these systems is needed and has yet to be written. Thankfully, repositories have begun to review their own users’ experiences. Most notably, Rachel Walton at Princeton University conducted an initial study of her repository’s site in 2017 with ten practical guidelines for improvements that could be used at any institution. But as repositories do not universally subscribe to UX thinking, let alone agree to standardize the look and feel of their sites for the benefit of their users, navigating these individual pages is cumbersome. And, without a national archival discovery system, cross-institutional discovery is further hindered.

Create a National Archival Discovery System to Improve Archival Literacy

Archival literacy, also called primary source literacy, allows a researcher to “accurately conceive of primary sources; locate primary sources; use a research question, evidence, and argumentation to advance a thesis; obtain guidance from archivists; demonstrate acculturation to archives; and follow publishing protocols.” Archival literacy standards focusing on the second goal—locating primary sources—attempt to remedy the problem of poor archival UX by teaching students and faculty workaround skills to optimize their searches based on current practices. While such pedagogy is necessary, it is not desirable. Archival literacy efforts should be able to prioritize understanding documents within their context rather than how to find them.

Archivists already emphasize document analysis over discovery. TeachArchives.org, created by Julie Golia and Robin Katz, represents best practices in the field of archival literacy as espoused by archivists. The website presents articles and exercises developed during a collaborative project between Students and Faculty in the Archives (SAFA) and the Brooklyn Historical Society that ran between 2011 and 2013. TeachArchives.org identifies four goals for instructors: designing specific learning objectives, selecting fewer materials, using small-group activities, and employing specific prompts. All four of these goals train instructors to think about how their students will interpret individual items rather than navigating an institution.

Subject specialists teach a broader understanding of archival literacy including both analysis and discovery. Sharon A. Weiner, Sammie Morris, and Lawrence J. Mykytiuk’s “The Archival Literacy Competencies for History Majors Checklist” from 2015 prescribes seven types of skills divided into 51 actions majors should be able to perform. The skill sets include: accurately conceiving of primary sources; locating primary sources; using a research question, evidence, and argumentation to advance a thesis; obtaining guidance from archivists; demonstrating acculturation to archives; following publication protocols; and advanced skills. Weiner, Morris, and Mykytiuk’s approach to archival literacy thus includes five units on engaging with a source critically and two predicated on discovery (locate primary sources and obtain guidance from archivists). While the subject specialists’ dual focus may result from the goal of teaching history majors how to conduct research, not just to close-read individual items, their inclusion of both objectives reflects the difficulty of searching for and within archives. Better UX practices within the modernist archive would allow instructors to focus on learning goals specific to their curricula rather than objectives based on an accurate assessment of the difficulties facing researchers.

Case Study: H.D.

On July 25, 2018, I searched for H.D. on ArchiveGrid. ArchiveGrid is the closest approximation to a national archival discovery system for the United States. These four screen captures show how difficult it is to parse an even well-established author’s archival footprint.


Fig. 1. Searching for H.D.

Fig. 1 shows my search for “H.D.”: the search generates a variety of results shown in what is called the “List View”; however, the first few results do not in fact refer to the poet. This is because (to simplify) ArchiveGrid is not controlled algorithmically like Google, which prioritizes the most frequently requested result, or Amazon, which tries to anticipate what you will find most useful based on what others found most helpful. Both approaches rely on robust linked open data, but linking data requires standardized metadata, which does not exist due to idiosyncratic descriptive standards employed over time across American archives with widely varying levels of financial and human resources.


Fig. 2. Summary view.

Archival literacy requires that we recognize how information is retrieved in order to understand the possibilities and limitations of discovery. While a system like ArchiveGrid aggregates finding aids related to a particular topic or individual, its results are driven by how well collections originally were described. H.D. is an example of a person whose holdings are both extensive and well described. Fig. 2 shows my switch to “Summary View.” This option immediately allowed me to narrow my search to the poet, “H.D. (Hilda Doolittle), 1886-1961.” In fig. 3, I clicked “H.D. (Hilda Doolittle), 1886-1961,” which refined my results within “Summary View” as seen in fig. 4.


Fig. 3. Refined results.

Clearly, H.D. is not the victim of archival silence as I was able to find what I wanted relatively quickly. According to ArchiveGrid, which does not claim to be a complete representation of all holdings in the United States, there are 38 separate collections which include content attributed to her. The final “Summary View” (fig. 4) represents H.D.’s archival footprint across people, groups, places, archives, archive locations, and topics.

Yet making sense of H.D.’s archival footprint requires more analysis on the part of the researcher. For example, which holdings are the most largest and therefore the most likely to be significant? Why did content arrive in these repositories rather than others? Are all of these holdings currently open to researchers? Are all of the collections equally well-described? Which hold content of a particular type, say H.D.’s letters?

The ArchiveGrid results are helpful, but the interface is not intuitive. Sophisticated researchers would be able to use their expertise to ascertain which results are of value to their particular project. Eventually, they would choose a hit representing one collection at one repository. ArchiveGrid would link that researcher directly to that repository’s digitized finding aid as it appears on the institution’s website. But a less sophisticated researcher is likely to get lost in the overwhelming amount of information. Without archival literacy skills, it is difficult to select which collection is likely to be the most useful; why collections are dispersed across a variety of institutions, including those that may not be represented in ArchiveGrid; and how H.D.’s rather large footprint shows that her holdings are relatively complete whereas an author with a smaller amount of material may either have more material still in private hands, in processing, or under restriction.

If even a straightforward search for a well-known author whose multiple collections are extensively described on a large discovery system can result in confusion, imagine what happens when one is searching collections related to people who are less well-known and thus are more likely to be the topic of fewer, less well described collections? Or whose content is held in institutions without the ability to make finding aids visible on ArchiveGrid or even general search engines like Google? These questions only begin to introduce how UX impacts the modernist archive.


Fig. 4. H.D.’s archival footprint.

Why UX Matters to Librarians, Archivists, and Researchers

As Weave’s panel of contributers observed back in 2015, UX rarely is seen as an integral facet of work in libraries and archives. Yet improving the UX for modernist archives would permit the correct identification of archival silences and enhance cross-institutional archival discovery thus supporting increased archival literacy. Meeting these goals would require a single, national archival discovery system that repositories could immediately populate with accession notes and later completed EAD finding aids.

Admittedly, this proposal is aspirational. While I argue that that archivists and UX librarians should continue to strive to meet these objectives, my primary intent is to show modernist researchers how structural constraints limit their archival discovery. Scholars should become more aware of how their discovery methods shape what they find and therefore what they may say. Recording these observations in publications would do much to heighten awareness regarding the impact of poor UX practices within the field of modernism.

After all, modernists consistently consider the value of the user interface. Modernists are sensitive to structure and form. The field continues to delicately dissect the way in which text and paratext inform one another in print as well as in the visual and performing arts. As modernism expands into the digital humanities, these considerations continue to be central. Early digital humanists debated what comprised a database and if a literary database could be simply be described as the digital content inhabiting an online repository or if the structure itself, the interface and its required code, needed to be included in the definition. More recently, modernist scholars have pondered how to apply the spatial humanities, often using Geographic Information Systems (GIS) tools, to better visualize their subject. Discussing UX is yet another way to engage, and understand, how what we understand of modernism is indelibly shaped by its information structure.

 

Notes

[1] Using the word “archive” to refer to these materials is problematic. Technically, archives are institutional records, not collections of personal papers. But as “archive” entered the critical lexicon in the humanities as a generic term for collections, especially when theorizing what these collections mean abstractly, I’ll also use the term archives to describe the broader network of papers. Furthermore, this blog post will only discuss holdings placed in American institutions as foreign repositories have myriad other circumstances influencing how easily their content may be discovered.

[2] More Product, Less Process (MPLP) attempted to solve the problem of extensive backlogs by relying on minimal description methods to push collections out to users. While this realistic solution to a massive problem caught on quickly, today it is now primarily used on institutional records (what the word “archives” actually means in technical parlance) rather than the personal papers which I argue make up the majority of what we think of when we think of the modernist archives. See: Mark A. Greene and Dennis Meissner, “More Product, Less Process: Revamping Traditional Archival Processing,” American Archivist 68, no. 2 (2005): 208-263.

[3] As extensive conversations around archival silences exist, those interested in archival silences resulting from non-creation, destruction, and non-acquisition should read one of the founding texts in those fields. I would suggest beginning with David Thomas, Simon Fowler, Valerie Johnson, and Anne J. Gilliland’s The Silence of the Archive (Chicago: ALA Neal-Schuman, 2017).

[4] Usability studies show that current EAD archival description practices still confuse researchers attempting to do comparative research, but without EAD, such discovery would not be possible at all. See Jihyun Kim, “EAD Encoding and Display: A Content Analysis,” Journal of Archival Organization 2, no. 3 (2004): 41–55 and Xiaomu Zhou, “Examining Search Functions of EAD Finding Aids Web Sites,” Journal of Archival Organization 4, nos. 3–4 (2007): 99–118.

[5] ArchivesSpace is an open-source, community-based software that allows archivists to avoid having to create the EAD coding themselves as the software completes this step automatically. Despite being open-source, ArchivesSpace is a subscription-based service that remains prohibitively expensive for many institutions. See: “FAQ” and  “Who’s Using ArchivesSpace.”

[6] ArchiveGrid harvests Machine-Readable Cataloging Records (MARC) from WorldCat, not EAD finding aids. MARC was designed to apply to bibliographic, rather than archival records, and dates to the 1960s. ArchiveGrid likely made this choice as MARC records existed prior to EAD records and because ArchiveGrid is a project of OCLC, which supports WorldCat. Archives West does not indicate how it populates its system while Arizona Archives Online and Online Archive of California integrate both EAD and MARC records and Rocky Mountain Online Archive, Texas Archival Resources Online, and Virginia Heritage use only EAD records.

Comments