Overview

Inclusive description is one of many terms used to describe DEI metadata practices, but what exactly does it mean? The dictionary definition of inclusion—the “I” in DEI—is the “practice of including and accommodating people who have historically been excluded (as because of their race, gender, sexuality, or ability).”^[1] Organizations such as Iowa State University have defined inclusion more broadly than the dictionary, as a “[c]ommitment to fostering a climate where all individuals have a sense of belonging through support and respect.”^[2] Inclusion thus refers to actions taken to respect and accommodate the needs of people—particularly those who have endured unequal treatment in mainstream society. Applying this idea to the information professionals’ understanding of description, inclusive description is information about a resource that not only helps users to discover and identify it but also meets those needs respectfully.^[3]

As a metadata professional, often working behind the scenes, you may not view yourself as having direct influence in the treatment of visitors to your library, archives, or museum.^[4] However, as a creator or enhancer of resource descriptions, you wield the power of language. Your work may be reactive, such as remediating biased or outdated terminology. It also may be proactive, such as writing descriptions that acknowledge the diverse backgrounds of your potential audiences. Either way, the metadata you produce is one of the primary ways that users interact with your institution. For that reason, you have a responsibility to foster respectful interactions.

This chapter provides recommendations for inclusive description in areas where metadata professionals have more flexibility and thus more responsibility to use accurate and respectful language. Free-text and uncontrolled elements—for instance, descriptions, summaries, biographical histories, notes, and alt text—offer a measure of descriptive freedom (see Figure 1.1). In addition, the title element, while usually restricted to the transcription of formal titles, offers flexibility when you need to devise titles for resources that lack them. (See Chapters 2, 3, and 5 for information about controlled fields.)

In any free-text element, references to human identity and personal names require attention and care. The next sections of this chapter offer strategies for describing identities and general tips about the use of names in free-text elements. However, sometimes metadata records contain problematic language that cannot easily be remediated. Insufficient resources and system limitations have meant that outdated metadata persists in many catalogs. Moreover, a commitment to historical accuracy may require the retention of offensive terminology. For these cases, the chapter concludes with sections on strategies for alerting users to potentially offensive descriptions: redaction or annotation; harmful language statements, also known as disclaimers or trigger warnings; and special genre terms that identify works as prejudicial.

References to identity

Identity is a complex concept that describes people’s character and personality, as well as their psychological orientation toward groups of associated people.^[5] A person’s identity comprises multiple, often intersectional, categories related to ability, age, class, gender, race and ethnicity, religion, and sexuality.^[6] Additional categories such as education level, family status, language, political beliefs, and work experience also contribute to identity.^[7]

Fundamental principles

If you are describing people in uncontrolled elements, when and how should you refer to identity? Consider the following principles to help you decide:

• Relevance
• Users’ needs
• Accuracy
• Respect

Note: These principles also apply to descriptions of people in controlled elements. See Chapters 2, 3, and 5 for guidance.

Relevance

First, prioritize relevance. Is identity-related information a significant aspect of the resource? Irrelevant references to personal characteristics may “other” individuals, or imply they are outside the norm.^[8] For instance, describing someone as a “woman scientist” could seem belittling. Moreover, gender may be unnecessary to mention, unless it is key to understanding or discovering the resource. Some identifying terms even could violate the privacy or preferences of the people you’re describing.^[9] (For more on othering, see Chapter 3. For more on privacy, see Chapter 2.)

Users’ needs

If identity is relevant and appropriate to mention, consider how identity-related language in a free-text element might help users to find, identify, select, obtain, and explore resources.^[10] In other words, would researchers use the terms in your description to search for and discover the resource? Would the terminology help them decide whether to access the resource? Would it provide context to enhance understanding of the resource?

Accuracy

When you do use identity-related language, always ensure accuracy, which is a longstanding principle of descriptive work.^[11] Record only factual information—that is, information found in the original resource or other authoritative sources. It is human nature to make assumptions about race, ethnicity, gender, or other identities, but you never should record your inferences without supporting evidence. For example, could a person potentially described as African American be from outside the United States or not of African heritage? Moreover, especially for living people, aim to follow personal preferences and privacy whenever possible. For instance, if an author has a personal website that states how they prefer to be identified, respect their wishes. (For more on these considerations, see Chapters 2 and 3.)

Respect

Finally, be respectful. Again, when possible, use the terminology preferred by the people and communities you’re describing. If your department or institution has developed relationships with representatives of local communities, ask them how they prefer to be described. Considering the constraints of the platform in use, explain what changes or additions are possible in your metadata. Be receptive to their feedback and thank them for their contributions, perhaps through a grant-funded donation or a credit on your website, or by offering access to information resources in exchange. If local outreach is not possible, consult with metadata professionals who have done similar work and refer to sources created or approved by the subjects of your descriptions. You also can refer to the resources below for terminology suggestions. (See the Introduction to this handbook for more considerations about outreach.)

Outreach is not a realistic option when describing people identified in historical materials created before the mid-twentieth century. In these cases, you may consult historians or other experts in the field, or you may rely on the materials themselves to determine which terminology was appropriate for the era. If you do reproduce outdated terminology in a title or free-text element, be sure to provide adequate context. (See Figure 1.2a.) Also consider applying a harmful language statement, using a genre term for prejudicial works, or both. (Refer to the sections on these strategies at the end of the chapter.)

Recommendations

Below are some specific recommendations for referring to identity, organized by several dimensions of diversity. Although this list was inspired by the work of numerous creators of inclusive metadata, consider it a starting point rather than a definitive standard.^[12] In every case, do research or conduct outreach to determine the terminology preferred by the described individual or community. (Refer the the “Respect” section above for tips.) Moreover, keep in mind these caveats:

• Members of any group may have conflicting preferences.
• Preferred terminology for any group may change over time.

Ability

• Be aware of differing terminology preferences in this area.
  • Person-first language, or describing a person before their disability, generally is considered respectful. For example, an individual with a disability.
  • However, some individuals and communities prefer identity-first language. In these cases, an adjective describing the disability may come first—for example, a hard of hearing person.^[13]
• Avoid condescending terms such as “handicapped,” “impaired,” and “challenged.”
• For terms to avoid and possible alternatives, consult the Cataloging Lab’s list of problematic Library of Congress Subject Headings (https://cataloginglab.org/problem-lcsh/)
• Consult the ADA National Network’s Guidelines for Writing about People with Disabilities (https://adata.org/factsheet/ADANN-writing)

Class

• Describe people equitably. When possible, specify names and other relevant details for all individuals described, not just the privileged and powerful.
• Avoid assumptions about socioeconomic status.
• Do not use dehumanizing terms like “the poor” or “the homeless.” Include the word “people”—for example, “poor people” or “people without housing.”

Gender and sexuality

• Do not assume gender or pronouns based on names or visual details. When gender is unknown or not relevant, use “person,” “individual,” “adult,” or “child” rather than “man,” “woman,” “boy,” or “girl.”
• When describing photographs, identify subjects from left to right rather than prioritizing by gender or status.
• Describe women by their full names, if known, rather than by their spouse’s name. If a first name is unknown, use the honorific and last name—for example, “Mrs. Smith” rather than “Mrs. John Smith” or “John Smith’s wife.”
• When describing individuals who identify as LGBTQ+, include information that you would include for cisgender heterosexual people—for example, marital, romantic, or parental status (if known and relevant).
• Consult these resources:
  • Trans Metadata Collective’s Metadata Best Practices for Trans and Gender Diverse Resources (https://doi.org/10.5281/zenodo.6686841)
  • Homosaurus, a linked data vocabulary of LGBTQ+ terms (https://homosaurus.org)
  • Cataloging Lab’s list of problematic Library of Congress Subject Headings (https://cataloginglab.org/problem-lcsh/)

Race and ethnicity

• Do not assume race or ethnicity based on names or visual details, without supporting evidence.
• Avoid presenting whiteness as the norm or focusing on white subjects at the expense of people of color.
• Consult these resources:
  • Archives for Black Lives in Philadelphia’s Anti-Racist Description Resources (https://archivesforblacklives.files.wordpress.com/2020/11/ardr_202010.pdf)
  • Creating Subject Headings for Indigenous Topics: A Culturally Respectful Guide (https://ir.library.oregonstate.edu/concern/defaults/8c97kz40m)
  • Cataloging Lab’s list of problematic Library of Congress Subject Headings (https://cataloginglab.org/problem-lcsh/) and Best Practices in Authority Work Relating to Indigenous Nations in the U.S. (https://cataloginglab.org/best-practices/)

Religion

• Do not assume religion based on race or ethnicity, or vice versa.
• Refer to Indigenous spiritual beliefs and practices as religion rather than mythology.

References to names

Names in devised titles, descriptions, and other uncontrolled fields differ from the authorized versions of names recorded in agent or subject elements or in MARC access points. First, names given in these fields are unstructured: they are not tagged or identified as names. Second, they are typically in direct order, rather than inverted. Third, they may include terms of address or honorific titles, pseudonyms, nicknames, spouses’ names, former names, and chosen or preferred names. This flexibility means that metadata creators must exercise careful judgment when determining the form of a name to record in an uncontrolled field.

Content or input standards, including Resource Description and Access (RDA) and Describing Archives: A Content Standard (DACS), direct metadata creators to include names in devised titles to indicate the subject of the resource or the name of the creator.^[14] In devised titles, these standards generally advise using the authorized version of the name (but in “natural language” or direct order). If no authority record exists, they advise using the form of the name used in the resource being described, the form given by other sources most frequently or most recently, or the form by which the person or entity is generally known.^[15]

It is good practice to follow these guidelines when including names in free-text fields. However, the flexibility offered by uncontrolled fields means that you also can provide context when needed. While giving precedence to preferred names, you can give alternative names and supporting information when warranted, particularly in description or note fields. If the resource itself uses an offensive or former name, you may be able to offer historical context in an explanatory note, or you may choose to add a harmful language statement. For more information about names, see Chapter 2.

Examples

In Figure 1.2a, the title (245), summary (520), and subject (600) fields use the name preferred by Sojourner Truth, but the summary also includes Truth’s former name. This assigned name provides historical context and could possibly assist with discovery or selection. However, if you were creating or editing this record, you might consult the resource or other authoritative sources to determine whether Truth used or acknowledged that name.

It also would be a good idea to cite the source of this summary and place the cited text within quotation marks. Moreover, you should consider describing Truth with the adjective “enslaved” rather than identifying her by the noun “slave.” For more information, see the source notes in the Library of Congress subject heading (LCSH) Enslaved persons,^[16] as well as the Redaction and Annotation sections below.

Figure 1.2b shows three MARC fields for a hypothetical book written by a trans person before their public transition, where the title page shows the author’s deadname (a name given at birth that they no longer use). Catalogers typically transcribe the name on the title page in the statement of responsibility (MARC 245 $c), even if it differs from the author’s name in their authority record, shown in the main entry—personal name field (MARC 100 $a).^[17] If you omit a deadname, you may add a note (MARC 500) about the change to the statement of responsibility. (For more on omitting and annotating data, refer to the Redaction and Annotation sections below. For more about names, authority records, and creator characteristics, see Chapters 2 and 3.)

In Figure 1.3, the devised title of a letter regarding the first African American elected to Phi Beta Kappa in Iowa, the subject’s preferred name, S. Joe Brown, is indicated in the resource and in other materials in the National Bar Association digital collection.^[18] Although the subject field uses the form of the name used in Drake University’s archival management system, the devised title uses this preferred name.

In Figure 1.4, the honorific title and name “Mrs. Welch” is retained in the transcribed title of an 1884 cookbook, but the name in the creator field is the Library of Congress Name Authority File (LCNAF) authorized form, which includes Welch’s full name and life dates.^[19] Moreover, the description provides additional context about the author by stating her professional credentials.^[20]

Figure 1.5 shows an example of a woman’s name that could have been hidden from discovery.^[21] The title of the source newspaper (appended to the transcribed title of the illustration) includes the publisher’s name, Frank Leslie. However, the description and publisher fields give additional context about the publication, specifically that Miriam Leslie, Frank’s widow, ran the paper after his death. Her name authority record uses Frank’s name rather than her own first name, but the description provides it.

Strategies for persistent problematic metadata or content

Realistically, your institution’s metadata will continue to contain outdated or biased language for several reasons. First, language continually evolves. Words meant as disparaging today may shift in meaning or even be reappropriated by targeted groups. Alternatively, words considered preferable today may become outdated or offensive in the future. Second, your institution may decide to retain certain terminology for accuracy and discovery. Third, the remediation of legacy descriptions takes time and resources, which may be in short supply.

Moreover, your institution’s resources likely will continue to contain problematic content, from titles containing offensive terminology to photographs depicting sensitive topics. Access and preservation are core tenets of libraries and archives. For that reason, these institutions always will hold a wide range of materials, including those with language and ideas that have become outdated or offensive over time.

Strategies for addressing problematic descriptions or content at the item level include redaction, editing, annotation, and access restrictions. Strategies that can be applied more broadly are harmful language statements and problematic genre/form terms.

Redaction

When text transcribed from a resource contains offensive terminology, redaction is one potential option, particularly for titles and summaries. Redaction is the removal or masking of letters or words—for example, by replacing them with punctuation marks (usually asterisks or dashes) or alternative text. In traditional cataloging, content standards such as RDA may indicate that alternative or non-transcribed text should be enclosed in brackets. In non-MARC metadata creation, such as for digital collections, brackets may not be required, depending on the content standard in use. Be sure to include any decisions about the use of punctuation in your local content standard or metadata creation guidelines.

Before deciding to redact text, be mindful that removing key words may impede discovery of the item. If your library services platform allows it, you may be able to separate what is indexed for search from what is displayed to the end user.^[22] (For more on display possibilities, see Chapter 5.) Moreover, library and archival professions generally oppose censorship. Be sure to consult with colleagues before enacting a redaction policy.

Figure 1.6 shows a redacted title for a digitized photograph with the caption “The Black Slave.”^[23] The metadata creator replaced the transcribed title with a devised title that uses updated language and provides context.

Figure 1.7 displays a redacted summary for a resource about a transgender person.^[24] The original summary for this resource contained insensitive language and used both male and female pronouns, despite the subject’s gender identity.^[25] In the redacted summary, the phrase “assigned female at birth” replaces the subject’s name at birth, and male pronouns are used consistently. In addition, the original final phrase, which characterized the subject’s life as a large-scale “deception,” was recast to focus on the media’s role in sensationalizing his story.^[26] If the book itself contains insensitive language or misgenders the subject, you also could add another note describing the issues, as in Figure 1.2b.

Annotation

Notes can be used to explain your approach to metadata remediation while respecting users’ needs. Summary and note elements have many potential uses, as evidenced by the long list of MARC note fields and the flexibility of the Dublin Core description element.^[27] You can use notes to indicate where potentially offensive language originated and explain what changes you made. If your institution has determined that access to the resource is restricted for privacy or sensitivity reasons, a note can explain how to access the resource. For transparency, you also may give a reason for the restriction. For example, in Figure 1.8, a note in the usage rights field helps users understand both why the video is restricted and how to access it.^[28]

Harmful language statements

Harmful language statements offer a means of alerting users to potentially offensive terminology. When these statements refer to the content of a resource as well as its description, they also may be known as content warnings, trigger warnings, or disclaimers. Because metadata remediation is an ongoing process, such statements acknowledge that the institution is aware of the issue and working to address it. In addition to alerting users, some institutions use these statements to invite users to report harmful language they come across in resource descriptions. A reporting feature can help you to target your remediation work. It also can promote positive engagement with users.

Harmful language statements can be displayed in item-level or collection-level metadata—for example, in a note element or field. Depending on the functionality of your library services platform, you also may be able to display a statement on the results or record pages in a discovery layer. For instance, libraries using Ex Libris Primo can implement a normalization rule that adds a statement to the full record display.^[29] In addition, statements can be displayed on the homepage of your library, repository, or collection website. However, be aware that many users go directly to search results, so they may be likely to miss homepage content.

Figure 1.9 lists harmful language statements and content warnings published by dLOC, DPLA, NARA, Iowa State University Library, and the National Library of Australia. dLOC’s statement appears at the bottom of its homepage and every results page. The DPLA and NARA pages include instructions on how to report potentially harmful language in archival descriptions. Iowa State University Library’s statement appears on the Digital Collections homepage and in a disclaimer element in the metadata for objects identified for DEI metadata remediation.^[35] National Library of Australia’s statement appears on the homepage, while a separate land acknowledgment statement is displayed on every record page in its online catalog.^[36] For more examples, refer to the Cataloging Lab’s List of Statements on Bias in Library and Archives Description (https://cataloginglab.org/list-of-statements-on-bias-in-library-and-archives-description/).

Genre/form terms

Another option for alerting users to problematic content is the use of genre or form terms, which describe the category of content or style represented by a resource. (Take care not to confuse genre terms with subject terms, which describe what a resource is about. See Chapter 5.) Depending on the metadata schema and application profile you use, you may be able to apply genre terms in these fields or elements: MARC 655, MODS genre, DPLA subtype, Dublin Core type, or VRA Core Work Type.

The Library of Congress Genre/Form Terms (LCGFT) vocabulary contains several terms that can be applied to works with clear biases.^[37] The Rare Books and Manuscripts Section (RBMS) Controlled Vocabulary for Rare Materials Cataloging (CVRMC) also includes terms for characterizing biased works.^[38] Figure 1.10 lists a selection of these terms. To facilitate the discovery and use of genre terms for prejudicial works, the Prejudicial Materials Working Group, part of the RBMS Controlled Vocabularies Editorial Group, compiled a list of new and revised terms in 2023.^[39] These terms are available through the Library of Congress Linked Data Service, which continues to be updated.

Because the decision to describe an entire work as biased is subjective, it is important to establish clear guidelines, preferably in collaboration with colleagues and stakeholders, for the consistent application of genre terms. The RBMS CVRMC offers some guidance in its definitions. It states that prejudicial works “exhibit bias in relation to a particular group or groups of people based on religion, race, ethnicity, gender, sexuality, ability, creed, national origin, etc.,” and that hate works “express hatred or advocate harm toward a particular group of people” based on the same dimensions of diversity.^[47] Again, however, consider not only the accuracy but also the utility of genre terms—will they help users to find, identify, select, and understand the resources described?

Conclusion

This chapter covered the importance of monitoring free-text or uncontrolled metadata elements for inclusive description, particularly when such elements contain references to human identity or personal names. The chapter highlighted the fundamental descriptive principles of relevance, users’ needs, accuracy, and respect and provided recommendations for several dimensions of diversity. It also gave examples of metadata containing various forms of personal names (former, preferred, authorized), as well as notes to acknowledge privacy concerns and historical context. For persistent problematic metadata or content, the chapter offered several strategies for alerting users: the redaction or annotation of metadata, the application of harmful language statements to individual records or entire collections, and the use of genre terms to identify works as biased.