The Library of Congress >> Especially
for Librarians and Archivists >> StandardsThe Library of Congress >> Especially
for Librarians and Archivists >> Standards
HOME >> MARC Development >> Discussion Paper List
HARMFUL LANGUAGE ADVISORY: In order to demonstrate the need for the proposed subfield, this discussion paper contains historical language that is offensive and harmful (primarily racist terminology in title statements in the examples).
DATE: December 18, 2024
REVISED:
NAME: Defining a New Subfield for Context of Transcribed Title Statement in Field 245 of the MARC 21 Bibliographic Format
SOURCE: The Bibliographic Standards Committee (BSC) of the Rare Book and Manuscripts Section (RBMS) of the Association of College and Research Libraries (ACRL), a division of ALA (American Library Association); co-sponsored by the Program for Cooperative Cataloging Advisory Committee on Equity, Diversity, Inclusion, Belonging, and Accessibility (PCC EDIBA)
SUMMARY: This paper proposes adding an optional subfield $z for "Title statement context note" to field 245 (Title Statement) to meet the reparative description need for immediate contextualization of racist, homophobic, ableist, and other biased, prejudicial, or hateful language found in transcribed titles of many special collections resources.
KEYWORDS: Field 245 (BD); Title statement (BD); Subfield $z, in field 245 (BD); Title statement context note (BD)
RELATED: 2024-DP02
STATUS/COMMENTS:
12/18/24 – Made available to the MARC community for discussion.
01/29/25 – Results of MARC Advisory Committee discussion: There was general support from MAC, with appreciation for the refinements over the previous discussion paper. There was consensus that the proposed subfield $z should be positioned at the end of the field, although there were reservations expressed about recording this data in brackets. There was also general consensus that the terminology deployed in $z should be left to best practices guidance from the respective cataloging communities, although a label and definition for the $z will still be needed in MARC. The use of $7 for data provenance was raised as an existing alternative solution, although the ability of systems to leverage this was questioned, particularly their ability to selectively use it in displays solely for this type of situation. The paper will return as a proposal.
This discussion paper, originally entitled "Adding Subfield $i to Field 245," was first presented at the Midwinter 2024 MAC meeting as 2024-DP02. During the discussion it became clear that the paper did not adequately convey the unique purpose of the new subfield, nor did it enumerate the many options that were considered before coming to the conclusion that adding a subfield to field 245 offered an appropriately simple and speedy solution to a pressing need in reparative description. A straw poll of attendees at the meeting revealed strong support for the concept, and the committee voted in favor of inviting resubmission.
We wish to clarify that the issue is not simply one of recording metadata provenance. Concerned members of the special collections community, particularly those in public services, brought a specific problem to the Diversity, Equity, and Inclusion Steering Group of the RBMS Bibliographic Standards Committee. Namely, the transcribed titles that form the bedrock of special collections cataloging can unintentionally act as gatekeepers when they appear in search results. This is especially true for titles of prints, photographs, ephemera, and other resources where devised titles could reasonably be expected. Users familiar with bibliographic conventions know that racist, homophobic, ableist, and other biased, prejudicial, or hateful language, referred to as "harmful language" henceforward, in an unbracketed title statement represents the resource in its own words, not the words of library staff. Note that the use of square brackets to identify supplied and devised information in Title Statements is required in Descriptive Cataloging of Rare Materials (RDA Edition) (release 2022.1.1; hereafter identified as DCRMR). Complaints from users reveal that this convention is not widely understood.
For many users, the online catalog is the first point of encounter with library collections. This is especially true for special collections materials, which users cannot physically browse in the library. Understood in the context of remote research and the need to page materials in closed stacks, the online catalog becomes one of the most public-facing elements of library services for special collections. In keeping with the ACRL Code of Ethics for Special Collections Librarians' mandate to "strive to improve collections access for all users" (emphasis ours), it behooves us to consider the online catalog as part of this commitment to equitable access. We must treat how materials are presented in our online catalogs with great seriousness if we are to ensure that users from diverse backgrounds are not turned away before they even reach the reading room. Although this paper focuses on applications to special collections cataloging for these reasons, users encounter transcribed title information in catalog records created according to a variety of descriptive standards, and the proposed subfield is not restricted to special collections applications.
This subfield will give catalogers the option to integrate contextual information at the point of encounter. It is not intended to warn users about harmful language, but to explain its presence. Broadly speaking, catalogers can identify titles appropriate for immediate contextualization based on the definitions of "Prejudicial works" and "Hate works" in the RBMS Controlled Vocabulary for Rare Materials Cataloging. That is, harmful language is language that "exhibit[s] bias in relation to a particular group or groups of people based on religion, race, ethnicity, gender, sexuality, ability, creed, national origin, etc." or language that "express[es] hatred or advocate[s] harm toward a particular group of people."
We acknowledge that harmful language can appear anywhere in a full catalog record, and that methods to contextualize that language within the record already exist. This paper addresses the specific need for contextualization within the MARC 245 Title Statement, the one field guaranteed always to be present outside the full catalog record when that record is represented elsewhere (e.g., brief search results, exhibition shortlists, digital image captions).
In addition, the writing group changed the proposed subfield from subfield $i, as previously seen in 2024-DP02, to subfield $z when revising the paper. During the discussion, there had been concerns that a 245 subfield $i would be used more broadly and flexibly by catalogers than was scoped in the original discussion paper (that is, strictly as context for a transcribed title in the case of harmful language). In response, we changed the proposed subfield to subfield $z, which is not otherwise used in the transcribed 24X, 25X, or 26X MARC Bibliographic fields, to underscore that the proposed subfield has a distinctive purpose. Finally, the subfield $i implicitly suggests that the subfield would be placed preceding the 245 $a, as it is used in the 246 and other fields, which would break the function of the indicators for non-filing letters. Institutions that do wish to display the content of the 245 subfield $z preceding the title can achieve that by programming their OPAC accordingly.
Field 245 is currently defined, in part, in the MARC 21 Bibliographic Format as follows:
Field Definition and Scope
Title and statement of responsibility area of the bibliographic description of a work.
Title Statement field consists of the title proper and may also contain the general material designation (medium), remainder of title, other title information, the remainder of the title page transcription, and statement(s) of responsibility. The title proper includes the short title and alternative title, the numerical designation of a part/section and the name of a part/section.
[For remainder of field definition and scope see https://www.loc.gov/marc/bibliographic/bd245.html]
- Indicators
- First - Title added entry
- 0 - No added entry
1 - Added entry- Second - Nonfiling characters
- 0 - No nonfiling characters
1-9 - Number of nonfiling characters- Subfield Codes
- $a - Title (NR)
- $b - Remainder of title (NR)
- $c - Statement of responsibility, etc. (NR)
- $f - Inclusive dates (NR)
- $g - Bulk dates (NR)
- $h - Medium (NR)
- $k - Form (R)
- $n - Number of part/section of a work (R)
- $p - Name of part/section of a work (R)
- $s - Version (NR)
- $6 - Linkage (NR)
- $7 - Data provenance (R)
- $8 - Field link and sequence number (R)
Accurate transcription is one of the fundamental principles of rare materials cataloging (DCRMR Introduction, section i.01.32). Scholars and librarians have long depended on it for identifying specific editions, distinguishing between similar resources, and collocating instances of the same resource across multiple institutions. Special collections are now expanding beyond this traditional scholarly audience, but being unable to provide immediate context for harmful language in transcribed titles hinders libraries in making the new audiences feel welcome.
For traditional users of special collections, the absence of square brackets means the same thing as the presence of quotation marks in the everyday world. That is, generations of scholars knew that unbracketed titles in library catalogs represent exact transcriptions from the resource itself, while square bracketed titles represent titles supplied from another source or devised by the cataloger. This convention is unfamiliar to new audiences in special collections, so a list of search results that presents prejudicial or hateful language without context unintentionally creates a hostile environment.
Bibliographic field 245 is, by far, the most prominent space where users encounter transcribed metadata. Not only is it the primary display in search results lists, it typically dominates the top of the record display. In addition, field 245 often represents the library resource in contexts outside the OPAC (for example, in internet search results, digital humanities data mining, and downloaded citations), and its content may be re-used – for example, mapped to Dublin Core as the title of an image, which may travel elsewhere on the internet and be re-used for a variety of purposes. Several community-based groups and guidelines make clear that best practice is to preserve but immediately contextualize creator-sourced original description containing harmful language. Examples include: the Anti-Racist Description Resources (October 2019), from the Anti-Racist Description Working Group of Archives for Black Lives in Philadelphia; The Metadata Best Practices for Trans and Gender Diverse Resources (version 2.5, November 7, 2023) from the Trans Metadata Collective; the Best Practices for Queer Metadata (version 1.2, July 15, 2024) from the Queer Metadata Collective, and the Protocols for Native American Archival Materials (April 9, 2007) from the First Archivists Circle. When the content of the 245 field travels outside of the context of the OPAC, contextualization present in other fields is lost. For these reasons, we believe it is appropriate and beneficial to embed the context in which a harmful title was created within the Title Statement itself. Placing this information only in a note, even a linked note, is insufficient.
Many of the standards for inclusive description that have been developed for special collections acknowledge the issue of harmful language in transcribed information. There is a tension between the need to represent collection materials as they present themselves (DCRMR Introduction, section i.01.322), while also adhering to the principle articulated in the Cataloguing Code of Ethics (updated September 2022) to approach standards "critically and advocate to make cataloguing more inclusive." Crucially, it is not currently possible to include contextual information at the point of contact with harmful titles. Some guidelines, such as the "Guidelines for Addressing Bias in Archival Description and Catalog Records" (Orbis Cascade Alliance, Unique and Local Content Program Unique Material in the SILS and Archives West Standing Groups, 2021), speculate that discovery layers and brief search results will one day be able to display specific 5XX notes contextualizing harmful transcribed language in bibliographic records, but we need a solution that will work now. This topic has also surfaced in inclusive description discussions and workshops including a presentation by Treshani Perera (Harmful Language in Transcribed Titles: A Case Study, at the OCLC Cataloging Community meeting, 2 February 2024) discussing both the earlier version of this presentation (from Less Talk, More Action: Adventures in Inclusive Metadata, ATLA webinar, October 10, 2023), and her own local practice of adding bracketed content warnings to title fields. This proposal responds to user comments and directly addresses a need expressed by catalogers working on inclusive description.
Straightforward techniques for qualifying or removing harmful language in the Title Statement field that were considered and rejected include the following.
2.2.1. Redacting problematic words
The established tradition of disguising words by replacing letters with asterisks or dashes fails to meet user needs in special collections for several reasons. For example:
- It can make a known title unrecognizable in search results.
- Even if a redacted title remains recognizable "by eye" to users doing known-item searches, redacting certain words or portions of words makes the title unrecognizable by machine. A user faced with a long list would not be able to find the particular title they want by keyword searching.
- Redaction distorts the historic record that special collections libraries seek to preserve and make available for study. Removing harmful language erases cultural realities. As Penguin Books Ltd. puts it in the Publisher's Note of some of their titles, "In this book are some expressions and depictions of prejudices that were commonplace at the time it was written. We are printing the book as it was originally published because to make changes would be the same as pretending these prejudices never existed."
- Redacting certain words or portions of words makes the title indistinguishable from titles issued with redactions (e.g., a print with the caption title The Conduct of the two B*****rs; image available from the British Museum Collection). Redaction also goes against the key rare materials cataloging principle of creating "accurate representations of the resource as it describes itself." (DCRMR Introduction, section i.01.322)
2.2.2. Enclosing the title in quotation marks
Using quotation marks to indicate that a title is directly transcribed from an item has the following problems:
- Special collections catalogers already use quotation marks to indicate that the resource itself has a title within quotation marks, something that often happens with pictures, especially caricatures.
- When used selectively, it would imply that the lack of quotation marks indicates a cataloger-devised title.
2.2.3. Devising a Title Statement and treating the transcribed title as a Varying Form of Title
Devising a Title Statement and moving the transcribed title to Bibliographic field 246 (Varying Form of Title) is appealing for its simplicity. Field 246 subfield $i already exists to provide context for the title that follows, and systems already index field 246 as a Title field. However, it also does not meet our needs:
- It shares the first three problems described in 2.2.1., above, by making known titles possibly unrecognizable by eye, definitely unrecognizable by machine, and by falsely sanitizing history.
- It assumes that the Title Statement and the Varying Form of Title will remain paired with each other, but that is only true within the OPAC environment. Image databases pull titles from field 245. Digital humanists collect historic title data for analysis from field 245. The richness of the original MARC record does not translate into other applications that draw on metadata about special collections material.
2.2.4. General note on source of title (500)
Special collections catalogers already make frequent use of the General Note field (500) to provide information about the source of the title in field 245. Indeed, such a note is mandatory when cataloging pictures according to Descriptive Cataloging of Rare Materials (Graphics). It does not adequately meet the need for context expressed in this paper for two key reasons:
- It does not appear at the point where the user initially encounters harmful content in Title Statements.
- It cannot easily travel with the title when it appears in other contexts (e.g., search engine results, digital image databases).
2.2.5. Source of title note (588)
In addition to sharing the problems identified with using a General Note (500) in 2.2.4., above:
- It would require redefining field 588 (Source of Description Note) because context for harmful language in title is not "information used for tracking and controlling the metadata" nor is it "especially of interest to other catalogers."
2.2.6. Harmful language advisory/Content warning
A broadcast advisory statement can be a useful option as a general reminder that library catalog records represent works as they present themselves, and that transcribed title information is important for conveying authentic history. However it does not meet the needs for providing context:
- Harmful content in the Title Statement has already been encountered, so it would be stating the obvious, not mitigating the situation.
- The generic nature of broadcast advisory statements makes them easy to overlook or dismiss.
Much of the recent development in MARC involves meeting cutting edge and future needs in a world of linked data and machine-actionable data. Adding a new free-text "Title statement context note" subfield may seem like a backward-looking solution. Code-based options that were considered and rejected include the following possibilities.
2.3.1. Subfield $7 - Data provenance
Conceptually, $7 is a promising solution. After all, providing context for harmful language in a transcribed Title Statement relates to the provenance of that statement – it originated with the creators and producers of the resource itself.
However, neither the $7's scope nor its implementation align with the proposed scope for $z. The $7 carries many kinds of encoded information including language, related manifestation, and source consulted, among others. In theory, eye-readable display capabilities could be developed based on particular codes, but this use of codes removes the flexibility and nuance that are important for contextualizing harmful language in titles. Additionally, the time and effort needed to create and maintain new codes and to configure systems to publicly display select – but not all – codes as eye-readable information prevents responsive implementation. Using free-text in the $7 alongside code values would create further complications, in that free-text is not readily machine-parsable for selective display. If $7 were to be displayed to the reader, additional information of many types and presentations would also be displayed – display of the subfield $7 would be an all or nothing presentation, including both codes (opaque and confusing to the user) and the free-text information contextualizing the title.
Ultimately, the audiences for the $z and $7 are different. As proposed, the subfield $z directly speaks to end users about the presence of harmful language, whereas information in the $7 is likely to be of direct interest to catalogers or to machine-actionable processes and would not be expected to directly display in the OPAC. Due to this difference, the subfields $z and $7 would complement each other; for example, $7 (dpeloe/dpsfz) eng could be used to indicate that the language of the note in subfield $z is English, or $7 (dpeaa/dpsfz) DFo to indicate that the Folger Shakespeare Library added the $z note.
2.3.2. Bibliographic field 883 - Metadata Provenance field with $8 field linkage
Instead of directly embedding Context of transcribed title in the Title Statement, control subfield $8 could be used to link the 245 (Title Statement) to an 883 ( Metadata Provenance) field. Currently, MARC field 883 is not intended for public display, but in theory the undefined second indicator could be defined to signal display or not. If context for a harmful transcribed title were provided in 883 $a, linked to the 245, and set to display, it would theoretically be possible for the context to display alongside titles in search results and record displays, similar to the way that $6 links different script representations to field 880, which can then be set to display alongside the primary parallel field.
Like approaches involving notes, using $8 field linkage does not ensure that the context remains associated with the title outside the MARC environment. It also shares the technical disadvantages found with using the $7 (Data provenance) subfield. Namely, displaying it publicly alongside the title in search results and record pages would require a great deal of planning and development. While it could ultimately be very useful to have the option of displaying Metadata Provenance alongside associated fields, the urgent DEI need to provide context for harmful transcribed titles requires a simpler solution.
This paper proposes using new subfield $z in field 245 to provide brief context for the source of a transcribed title statement containing harmful language.
In field 245 (Title Statement) in the MARC 21 Bibliographic Format, define subfield $z as follows:
$z - Title statement context note (NR)
Note providing context for the source of a title statement containing harmful language recorded in any transcribed element of the field. The note is written in a form that is adequate or intended for public display.
Note: Examples show a variety of possible wordings, and are formatted according to Descriptive Cataloging of Rare Materials practice, where information interpolated within a transcription field, including the General Material Designation if present, is enclosed within square brackets.
Example 1
245 10 $a Jicarilla Apache Brave and Squaw, lately wedded. Abiquiu Agency, New Mexico $h [graphic] / $c T.H. O'Sullivan, photo. $z [Title transcribed from item]
Example 2
245 10 $a Deaf & dumb children of St. Rita's School, Cincinnati, singing Star Spangled Banner $z [Title provided by publisher in 1918]
Example 3
245 10 $a Nigger war bride blues / $c words by Jimmie Marten, music by Mitch LeBlanc $z [Language advisory: title from cover]
Example 4
245 00 $a G-men, sheriff aides, hunt Jap spies Los Angeles, Calif. $z [Language advisory: title from 1942 newspaper caption]
Example 5
245 14 $a The origin of civilisation and the primitive condition of man : $b mental and social condition of savages / $c by Sir John Lubbock. $z [Title as given on title page]
Example 6
245 12 $a A list of American organizations for cripples / $c compiled by Douglas C. McMurtrie $z [Language advisory: title dates to 1916]
No special provisions anticipated. The BIBFRAME conversion programs can be modified to accommodate this change. The MARC-to-BIBFRAME conversion could place this note within the title resource. An example:
<bf:title>
<bf:Title>
<bflc:nonSortNum>4</bflc:nonSortNum>
<bf:mainTitle>
The origin of civilisation and the primitive condition of man
</bf:mainTitle>
<bf:subtitle>Mental and social condition of savages</bf:subtitle>
<bf:note>
</bf:Note>
<rdfs:label>[Title as given on title page]</rdfs:label>
</bf:Note>
</bf:note>
</bf:Title>
</bf:title>
6.1. Have we sufficiently demonstrated that harmful language in transcribed title statements is a special case in reparative description, distinct from harmful language elsewhere in the catalog record or in the resource itself?
6.2. Are there options other than the one presented that would meet the needs of reparative description in relation to title statements in special collections?
6.3. Should the wording of subfield $z be prescribed, partially prescribed (e.g., always start with "Language advisory"), or left to each cataloger's judgment?
6.4. Should subfield $z follow the 245 subfield $a; follow the subfield $h; or be at the end of the 245 field?
6.5. Are there any potential consequences that this paper does not address?
HOME >> MARC Development >> Discussion Paper List
| The Library of Congress >> Especially
for Librarians and Archivists >> Standards (04/24/2025) |
Legal | External Link Disclaimer | Contact Us |