That was the topic discussed recently by OCLC Research Library Partners metadata managers, initiated by Sharon Farnel of University of Alberta and Jennifer Baxmeyer of Princeton. The impetus for the discussion was the OCLC Research report on the findings of a 2017 RLP survey on equity, diversity, and inclusion. Of particular interest to metadata managers were the following:
- 55% of respondents have changed metadata descriptions in archival collections due to their institutions’ EDI goals and principles.
- Three of the top four areas that respondents plan to change in the future are:
- terminologies and/or vocabularies (52%)
- metadata descriptions in digital or digitized collections (58%)
- metadata descriptions in library catalogs (70%)
Metadata managers thought that archivists had more flexibility in applying vocabularies that are more appropriate and respectful to marginalized communities to their archival collections than librarians had for published materials. As archival collections are unique, there is less pressure to use the same vocabulary as others, while records in library catalogs for publications need to have the same standardized terminology for them to be copied and re-used by other libraries holding the same item. Archivists are more likely to be more closely connected to the donors of their collection and thus more sensitive to their concerns. Fewer metadata specialists have the opportunity to deal directly with the public they serve.
Our discussions also noted that collections themselves can embed biases, which are then reflected in the metadata describing them. Many institutions are now engaged in diversity and inclusion activities, with more inter-actions between metadata specialists and curators.
Most of our discussion focused on the complexity of embedding equity, diversity, and inclusion in controlled vocabularies used in library catalogs. Points raised:
- Using other, less-offensive vocabularies can split collections, thus hampering discovery of all relevant materials.
- The process of changing standard subject headings can be very slow.
- We doubt that there can ever be complete consensus about any given text string. Terms that may be offensive to one community may not always be clear. (One example of this: “Dissident art” rather than “Non-conformist art”.)
- Changing headings in legacy data can require a massive undertaking. Some noted it would be less labor-intensive to present a “cultural sensitivity” message as part of the search interface to alert users that terms and annotations they find in a catalog may reflect the creator’s attitude or the period in which the item was created and may be considered inappropriate today in some contexts.
- Some who have tried to use local vocabularies more suitable for their context and communities too burdensome to maintain and abandoned them. Other attempts such as the University of Manitoba Libraries’ creation of local authority records for indigenous people instead of “Indians of North America” cannot be shared with other systems.
- The language of our controlled vocabularies may be exclusive to audiences who do not read English. Ohio State University libraries has tried to address this by developing some non-Latin script equivalents of subject terms.
- Language codes can also be exclusive. Australia has hundreds of indigenous languages, all lumped together under an “Aus” language code. Lumping together disparate languages prevent surfacing materials in a specific language; the National Library of Australia has proposed separate MARC language codes for “aus” languages.
- Linked data may offer the flexibility of applying different “labels” associated with the same identifier that could differ according to the context and language of the community. Meanwhile, Traditional Knowledge (TK) Labels offer a “tool for indigenous communities to add existing local protocols for access and use to recorded cultural heritage that is digitally circulating outside community contexts.”
- Current classification systems are apt to “ghettoize” ethnic groups. Rather than include them as part of an overall concept like history, education, literature, they tend to be grouped together as one lump. As the need to shelve materials together has subsided as institutions store more publications off-site, there is less need to have just one classification in a record, but few apply multiple classifications in one record.
We already use multiple vocabularies in our library metadata. Libraries use Medical Subject Headings in addition to Library of Congress Subject Headings (LCSH); art libraries use the Getty Art and Architecture Thesaurus for the same items described by others applying LCSH. Perhaps use LCSH for broader descriptions and other vocabularies for granularity? The National Library of Australia plans to use the Australian Institute of Aboriginal and Torres Strait Islander Studies (AIATSIS) Thesauri for materials related to Australia’s indigenous peoples.
We acknowledge that we have implicit or hidden biases in our descriptive metadata as well. We may identify “African-American” images in photo collections but not “white” or Caucasian; Library of Congress Subject Headings only mention race when the person is not white (e.g., “Men” and “African-American men.”) Should we categorize all people? How can such categorization be objective?
We agree that consulting the communities reflected in our descriptive metadata and access points would help facilitate moving to terms that are more appropriate and respectful. Our metadata is currently created according to Western knowledge constructs, and our systems have been designed around them.
Several metadata managers also serve on the Digital Library Federation’s Cultural Assessment Metadata Subgroup which is currently conducting a literature review of this issue. One result of last year’s survey of the OCLC Research Library Partnership was a web page listing resources related to equity, diversity, and inclusion: EDI in the OCLC Research Library Partnership Survey – Resources. We welcome suggestions for additional resources!
Karen Smith-Yoshimura, senior program officer, works on topics related to creating and managing metadata with a focus on large research libraries and multilingual requirements.