Skip to content

Harmful Language

Definition

Language that causes uncomfort, pain, feelings of unsafety to an individual or group of people.

Definition source: Combatting Bias definition.

Stakes

part of: discrimination

Harmful language can cause hurt to people and make the spaces they access your data in feel unsafe.

Where does it occur in the lifecycle?

3 - Process

Questions to consider throughout your work

  • Does your data contain offensive/harmful language and/or categories?

Examples

Good-better-best practices

Good Better Best
Present offensive terms taken directly from the archive in quotation marks. Explaining that words were or were not maintained from (primary) sources and/or other datasets, and explaining why this decision was made.1 “For collections or discovery portals where users may be harmed by encountering offensive language, it may be appropriate to replace offensive language in primary user-facing description. For an example of this approach, refer to the language policy for the Find & Connect website on child welfare in Australia.”2
Contextualise potentially harmful descriptions and words. Acknowledge the evolving nature of terminology guidelines. During annotation, use alt- and pref-labels to curate how your dataset will be viewed and read.
Reference community-approved guidelines. See resources below. Create research guides with relevant search terms for users on offensive language in historical records. Create standardized institutional language that can be reused. See resources below.
Implement participatory methods for tagging and description. Consult communities and experts on terminology. See Collaboration.

Resources


  1. Adapted from Archives for Black Lives, Anti-Racist Description Resources (2019). 

  2. Taken from from Archives for Black Lives, Anti-Racist Description Resources (2019).