Arbiters of Ugliness: A Review of Strategies for Describing Arbiters of Ugliness: A Review of Strategies for Describing Offensive Archival Materials Offensive Archival Materials

As archivists increasingly concede that neutrality is impossible, we suggest that non-action is still action. It follows that to treat reasonably offensive records as any other record is to apply an interpretation that they are innocuous, unremarkable, and uncontroversial. Archivists may perceive the stakes of describing these materials as particularly high, but they lack a comprehensive set of descriptive strategies in consideration of interpretive ethics. As a result, existing practices are likely to be local or ad hoc. This research aims to identify and explore descriptive strategies archivists use which serve to construct (or concede) the meaning that certain historical materials are potentially offensive using a combination of literature review, evaluation of finding aids and descriptive metadata, and exploratory interviews with archivists and other memory institution professionals. Results supported the assumption that strategies are largely ad hoc practices and local norms influenced by a handful of culturally sensitive descriptive protocols; underscored that strategies are contextually implemented; and revealed practical and philosophical divisions between archivists working with largely offensive collections and those working with largely uncontroversial ones. These findings suggest that while a comprehensive set of descriptive strategies may support a community grappling with its professional legacy, there may be no strategy capable of reconciling matters of ethics and discoverability, and the highly contextual practices are incompatible with overly rigid frameworks.


Introduction
One is not born, but rather becomes, a woman (de Beauvoir, 1949).
So, too, a collection becomes.The moment it is deemed historically significant, it ceases to be filing cabinet contents or basement clutter and becomes instead a Collection.It absorbs further meaning the moments it is appraised, arranged, described, referenced, and revisited.
The traditional Enlightenment view of archivy as a science and archivists as neutral imagines the repository and its practitioners as non-participants in the construction of meaning, but a postmodern view posits that while "some of what makes a record meaningful is inscribed in it by those who literally made it...most of what makes a record intelligible lies outside its physical borders in its context of interpretation" (Nesmith, 2002).As such, the postmodern view holds that "no approach to archival description, no descriptive system or architecture, can escape the reality that it is a way of constructing knowledge through processes of inscription, mediation, and narrative" (Duff & Harris, 2002).This attitude is reflected in the field's ongoing reckoning with neutrality, ushered in by a postmodernist shift around the late 1980s.As archivists increasingly concede that neutrality is impossible, we suggest that non-action is still action.It follows that to treat reasonably offensive records as any other record is to apply an interpretation that they are innocuous, unremarkable, and uncontroversial.
Neutrality-motivated descriptive practices aim to extend users the opportunity to make their own evaluations.Conversely, describing materials according to their potential to offend risks constructing meaning on behalf of users who will internalize it, and deciding what constitutes offensive is a highly subjective, contentious, and political act imbued with a great deal of power and authority.If meaning making is inevitable, a postmodernist will not ask whether they ought to construct meaning but might grapple with what meaning they are willing to construct and the boundaries of their construction.Archivists may perceive these stakes to be particularly high when tasked with describing those things with great potential to offend, but they lack a comprehensive set of strategies to describe those materials in consideration of interpretive ethics.As a result, existing practices are likely to be local or ad hoc (Nelson, 2020).
The Information Warrior Journal Winter 2024, https://doi.org/10.22237/tiwj/1693353840https://digitalcommons.wayne.edu/tiwj/This research aims to identify and explore descriptive strategies archivists use that serve to construct (or concede) the meaning that certain historical materials are potentially offensive using a combination of literature, evaluation of finding aids and descriptive metadata, and exploratory interviews with archivists and other memory institution professionals.

Overview
Literature on archival management of potentially offensive materials is slim and often theoretical.The most comprehensive practice-based literature tends to suggest strategies for specific communities and subject matter (see: Archives for Black Lives in Philadelphia, 2020;ATSIDA, n.d.;ATSILIRN, 2012;First Archivist Circle, 2007).Rarely are broad sets of strategies presented in a way that emphasizes their multi-context potential.
Existing strategies were initially identified through this literature, while a review of finding aids and descriptive metadata suggested additional strategies.Adjacent literature on content warnings was reviewed, as well as content warnings and like statements issued by more than eighty repositories and units.
Strategic keywords were used to search local and aggregated databases for descriptive records that may show evidence of a unique descriptive treatment.Keywords ranged from the practical (e.g., "content warning," "offensive," "outdated") to the uncomfortable (e.g., slurs that are difficult to query, much less utter).Lack of consistent practice and the likelihood that certain practices wipe evidence from description means there is no one reliable route to discovery, and the subtlety of some practices makes it difficult to evaluate their purpose with certainty.
To evaluate additional and perceived strategies, unstructured interviews were held with five professionals whose duties include description.Interviewees included one in collections management; one in processing management; one in project management; one in digital collections; and one in audiovisual collections.Four worked at repositories affiliated with a university, and one at a government repository.Three worked in archival libraries, and two in museum archives and collections.Importantly, three interviewees worked at repositories with broad collecting scopes where they sporadically encounter offensive materials among otherwise uncontroversial collections; two worked almost exclusively with offensive materials due to their institutions' collecting scopes (e.g., war crimes, slavery, genocide, etc.).
The Information Warrior Journal Winter 2024, https://doi.org/10.22237/tiwj/1693353840https://digitalcommons.wayne.edu/tiwj/While offense is socially, geographically, historically, and temporally subjective, some definitional guideposts were sought to help focus interviews on the more deeply offensive materials (Jones, 2010).Ruth Ann Jones, a special collections librarian at Michigan State University, suggests four categories that we might use to differentiate between different varieties of "controversial" materials: 1. Material that is offensive or abhorrent to almost everyone: hate literature, Holocaust denial literature, child pornography.
2. Material where opinion is divided -it's offensive to some but not all: pornography featuring & intended for adults; political opinions opposite to one's own; any opinions or beliefs that are hotly debated in the public area; anything related to gay/lesbian/bisexual/ transgender people; ethnic humor.

Material promoting dangerous or illegal activities. The Anarchist Cookbook. Final Exit: The
Practicalities of Self-Deliverance and Assisted Suicide for the Dying.
4. Material that some people may not consider appropriate in public, even if they enjoy it in private or among friends: bathroom humor; crude language, etc. ( 2010).
The materials discovered in this research exhibiting descriptive recognition of their potential to offend were overwhelmingly racist and antisemitic in nature.Occasionally, materials falling into Jones' second category of divided opinion were found to include descriptive recognition.These examples generally favor leftleaning values (e.g., content warnings for anti-abortion, anti-gay, and fundamentalist Christian rhetoric), arguably reflective of progressive liberalism's influence on librarianship (Knox, 2020).Like any definition, Jones's categories are restricted to the temporal and spatial context of their development, underscoring how moot a task it is to qualify offense.For these purposes, the materials under consideration best meet Jones' criteria for broadly offensive, with some inevitable crossover into divided opinion due to their shared social and identitarian nature.
Lastly, different materials may warrant different adjectives than "offensive," as it will sometimes unfairly suggest overreaction or hypersensitivity."Controversial," "harmful," "objectionable," "sensitive," "controversial," and "outdated" are all additional terms invoked for roughly the same concerns.For clarity and consistency, "offensive" is the term relied on here.Its imperfection appropriately mirrors some of the The Information Warrior Journal Winter 2024, https://doi.org/10.22237/tiwj/1693353840https://digitalcommons.wayne.edu/tiwj/descriptive challenges assessed herein.The choice also conveys the motivations underpinning this research more transparently, which are distinct from (but not incompatible with) reparative description.While practices considered here have reparative potential, this work springs from different (but not incompatible) motivations to protect access to materials that are tremendously valuable precisely because they are offensive from deliberate and functional censorship.If these materials are worthy of preservation, they are worthy of discovery, which certain reparative practices may impede.

Punctuation, substitution, and markup strategies
Among the most identifiable strategies is the use of punctuation to denote added language or signal offensive original language.These included the addition of modern language equivalents in square brackets and placing quotation marks or inverted commas around original language (Chilcott, 2019).This method functions to alleviate searchability issues associated with outdated and offensive terms by replacing them with terms users are more likely to query.While archivists have traditionally described materials using the language of the materials themselves, postmodern practice accommodates additions and substitutions in service of searchability.
No examples were discovered wherein a modern equivalent was obviously replaced (by sole use of a bracketed term), but untraceable substitutions for offensive language are likely made to the same extent that more searchable or descriptive substitutions are made without necessarily indicating so.Just as this practice impeded discovery for this research, one interviewee observed that the practice obstructs the pursuit of legitimate research questions, such as the evolution of slurs.It also collapses more differentiating terms, like exchanging the word "negro" for "black," which instantly buries a record among others sharing a more common and not strictly synonymous word.
While no interviewee expressed support for untraceable replacements, three employed modifying strategies, arguing they serve the dual purpose of retaining original language and implicitly communicating an institutional value of cultural sensitivity to users.Two interviewees opposed modifying strategies entirely, calling them "editorializing."An example is a collection of railroad tickets earmarked for enslaved passengers and emblazoned "NEGRO TICKET."Although the objects' own language is repeated on the finding aid, added quotation marks indicate the language is precisely what the materials call themselves.Arguably, it also The Information Warrior Journal Winter 2024, https://doi.org/10.22237/tiwj/1693353840https://digitalcommons.wayne.edu/tiwj/functions to distance the repository from the language, particularly in the narrative where the phrase "socalled" precedes the original language.Seemingly innocuous characters still do rhetorical work.
Likening it to restoration and colorization that fundamentally "change what the thing is," one interviewee asserted that the addition of quotation marks, inverted commas, or bracketed alternatives interfere with authenticity.Notably, both interviewees who opposed the practice work with almost strictly offensive materials, expressing greater concerns with authenticity.These practitioners favored a "what the eyes see" approach due to their subject areas' vulnerability for revisionism and unsubstantiated denial of historic events.Two strategies were identified that mitigate ethical and interpretive issues associated with replacement and other modifying strategies.
Amanda Gailey describes a markup strategy to faithfully transcribe Joel Chandler Harris stories, while modern language equivalents and regularized spellings sit searchable beneath the public interface.Best known for his Uncle Remus and Br'er Rabbit stories, Harris signals anti-Black tropes like laziness and ignorance by phonetically exaggerating a southern Black dialect."Brother rabbit" might instead be represented as "br'er rabs," which obscures the transcribed works from data harvesters and search engines.However, regularizing the text to alleviate this problem alters meaning by subduing offensive nature and "fundamentally undermines Harris's artistic project" (2010).To reconcile this conflict, Gailey and colleagues used XML markup to encode both verbatim and regularized transcriptions, differentiating using <orig> and <reg> tags and maintaining relationships between lines by nesting them both in the <choice> tag.Original transcription is the default public view, preserving users' ability to make their own first impressions, while regularized transcription improves discoverability in the background.
Another potential alternative to replacement was observed on the record for a lynching photograph.
The item includes a handwritten caption that could function as a moderately descriptive, supplied title.However, the caption invokes a flippant euphemism for lynching, "necktie party."A devised alternative was placed in the title field, while the original caption was transcribed in the notes field.While the devised title was no more descriptive than the complete original caption, the victim's name was added in a seemingly symbolic act to rehumanize a subject dehumanized by both their manner of death and the camera that captured it.
Although we can only speculate the cataloger's motivation for devising a title, simply relegating the original The Information Warrior Journal Winter 2024, https://doi.org/10.22237/tiwj/1693353840https://digitalcommons.wayne.edu/tiwj/caption to a less consequential field may function to impugn the offensive language while retaining it for searchability, context, and transparency.

Controlled vocabularies and critical subject indexing
A practice known as critical subject indexing involves making subject and keyword choices in consideration of classification's role in maintaining systems of oppression (Bruce, n.d.).Three interviewees reported opting for local alternatives where authoritative vocabularies fell short, all invoking the same example of "Enslaved persons" as an alternative to "Slaves," which was observed on numerous finding aids before the Library of Congress adopted the same revision.All three also cited the same set of recommendations for writing and teaching about slavery as influencing that particular choice.Gabrielle Foreman, et al. write, "using enslaved (as an adjective) rather than 'slave' (as a noun) disaggregates the condition of being enslaved with the status of 'being' a slave.People weren't slaves; they were enslaved" (n.d.).
Critical subject indexing may also involve deliberate choices between different subjects available within a controlled vocabulary; Kate Holterhoff argues that opting for the Library of Congress Subject Heading "Racism" instead of "Race" more accurately describes visual materials depicting violence and exploitation of Africans by colonists (2017)."Racism" is more critical and descriptive, and the choice suggests a more sophisticated visual literacy and knowledge of the subject matter.
One interviewee acknowledged that users may still search out-of-favor terms and had explored the ability to invisibly tie potential variants to roughly synonymous subjects.This strategy is comparable to Gailey's use of XML, except the locally preferred terms would be displayed in the public interface with outdated equivalents tied to them in the background.For example, a search for "Slaves" would return results for records indexed with the subject "Enslaved persons" by treating them as synonyms in the backend.This enables archivists to remove offensive language while preserving a discovery path using language that has not fallen entirely out of use.While using this strategy to bury important original language raises ethical concerns, it is appropriate to revise subject headings and other archivist-applied language.
These choices are rhetorical in that they contend with unwanted connotation, but they are also symbolic and example setting.It is implied that language choices have socially transformative potential, and The Information Warrior Journal Winter 2024, https://doi.org/10.22237/tiwj/1693353840https://digitalcommons.wayne.edu/tiwj/While the Peabody provides resources outside of the database that better contextualize the Zealy daguerreotypes, there is no added description to help contextualize them except for a note about historical language that is universally applied to all database objects regardless of offensive potential.When asked about potential for critical description to unduly influence user interpretation, one interviewee replied, "to be timid and avoidant is an interpretation."The omission of "Enslaved persons" or "Scientific racism" subjects may function to interpret the Zealy daguerreotypes more deeply than the inclusion of "Portrait" and "Man." It is worth noting that repositories often impose access restrictions, especially digitally.Harvard no longer restricts reproduction and publication of the images, but the Peabody omits five images from the database records in which subjects are photographed from the waist down (Barbash, 2020).This is also common practice with lynching photography.One interviewee who manages a strictly offensive collection explained that the repository had opted against a publicly searchable artifact database to eliminate the potential for images to be irresponsibly saved, reproduced, and circulated.At the time, the repository provided contextualizing articles with select low-resolution images on their website and limited access to collections primarily to in-person researchers.While access begins with competent description, a well-described record need not equal ease of access.

Content warnings and acknowledgement statements
Of the strategies observed, content warnings most clearly anticipate patron objection to archival materials, in effect stamping them "offensive."They vary rhetorically, as well as in their practical application and universality.The most broadly applied statements can be found on repository websites divorced from collections databases, which are necessarily general.When they appear within finding aids and descriptive records, they may be non-specific and universally applied to all records, or they may be applied and tailored case-by-case.Finally, they are found beneath many different headings.To name a few, whether a statement is named a "warning," "acknowledgement," or "disclaimer," each implies different institutional attitudes and motivations for issuing them.
Standardized statements that are applied universally to all descriptive records or posted outside of collection databases set the expectation that offensive materials are more norm than exception in historical collections.They also relieve archivists from acting as authorities on what counts as offensive.However, The Information Warrior Journal Winter 2024, https://doi.org/10.22237/tiwj/1693353840https://digitalcommons.wayne.edu/tiwj/standardized statements are necessarily non-specific, which can result in more value-laden language compared to tailored statements.One standardized example states, "there are materials in our collections that may be offensive or harmful, containing racist, sexist, Eurocentric, ableist or homophobic language or depictions" (Washington State University, n.d.).Because they address materials broadly, standardized statements like this one rely on more judgmental adjectives compared to collection and item-level statements.At lower levels, archivists can simply call something by its name: a swastika is the name for the symbol whether we disapprove of antisemitism, and blackface is what we call white people painted black whether we deem it racist.However, headings such as "content warning" risk influencing the way readers interpret the matter-of-factness that may follow.Additionally, broader units of materials can end up descriptively treated as if they are about their singular offensive outliers.
Content warnings' contentiousness in broader contexts may be attributable to the same interpretive intervention that gives some archivists pause.Asked about potential for users' interpretation to be influenced or warnings internalized, two interviewees who encounter offensive materials sporadically echoed postmodern tenets that interpretation is inevitable.Both argued that interpretive restraint is outweighed by harm mitigated through content warnings that empower users to proceed or turn back.While this is a common justification for issuing warnings, some research suggests content warnings are either "functionally inert or cause small adverse side effects" in individuals with trauma history (Jones et al., 2020).
That finding matches a concern shared by two other interviewees, both managing largely offensive collections, that content warnings do not accomplish what they intend.While both identified the purpose of warnings as to protect members of certain affected communities from trauma responses, both doubted that they accomplish that.One worried they represent more of an escape hatch for others to avoid uncomfortable histories in which their ancestors may be implicated.The other reported experiencing content warnings as selffulfilling prophecies, observing that patrons generally rise to the occasion when they have been prepared to feel outrage or distress prior to collection tours.Other research suggests that while content warnings do not temper negative reactions to certain images, they also do not prompt more negative interpretations than the same materials without a warning (Bridgland et al., 2019).
The Information Warrior Journal Winter 2024, https://doi.org/10.22237/tiwj/1693353840https://digitalcommons.wayne.edu/tiwj/The same interviewee also sensed that the need to protect affected communities is inflated, using slurs as an example.Warnings may be purely symbolic for members of certain communities who are frequently the subject of slurs.This inflation may be the consequence of an expanding definition of the word "trauma," which is now used to describe more diffuse varieties of trauma than in the past.While content and trigger warnings were previously familiar to protect war veterans and survivors of sexual assault whose trauma responses may be acute, they are now commonplace for more routine experiences of harm.
Both interviewees who worked with almost entirely offensive collections were strongly opposed to content warnings, and both indicated that their repositories or collections were named in such a way that they functioned to warn by clearly signaling a difficult scope.They also both remained reluctant of warnings by a different name (e.g., statements, notes, disclaimers), and one continued to express hesitation when asked whether they would consider using them at repositories where users are more likely to be surprised by offensive materials.

Conclusion, or How many archivists does it take to screw in a lightbulb?
It depends.This joke, recited by an interviewee, typifies archival practice: professional norms and principles are often a poor fit for decisions that confront us in practice, and we consider a variety of contextual factors to inform those decisions.We fill in using what we know about our profession and ethical code; our repository and its community, policies, norms, and resources; ourselves, our biases, motivations, and authority to interpret; and the records and artifacts that cross our desks, their content, and context.How to descriptively manage offensive materials is firmly among those "it depends" decisions, which archivists allow to be shaped by context.This situational decision making is demonstrated by the most notable finding that interviewees' attitudes and practices were neatly divided according to whether a.) the practitioner strictly managed offensive collections or b.) sporadically encountered offensive materials among uncontroversial collections.All interviewees describe their practices as "calling it what it is," and all nested their descriptive choices in their intention to act ethically and honestly; however, the two groups conceptualized "calling it what it is"