What is the Equity & Bias Language Analysis?
Reviewing class specs for inclusive language is a time-consuming task, and one that often gets deferred because there's no efficient way to know where to start. Holly's Equity & Bias Language Analysis does the first pass for you. It reads a spec the way an equity reviewer would and flags the specific language that raises concerns, along with a plain-language explanation and a suggested fix.
The analysis evaluates specs across four dimensions:
Gender Language: Gender-specific terms where neutral alternatives exist, such as "foreman" or "he/she" constructions.
Bias Language: Terms that may exclude or disadvantage certain groups.
Accessibility: Language and readability concerns that make a spec harder to understand or that don't support reasonable accommodation.
Fair Requirements: Requirements that may be stricter than the job demands, which can screen out qualified candidates.
Note: This analysis uses AI to identify language issues and suggest replacements. AI-generated findings support your team's judgment; they aren't a legal determination or a substitute for human review. Always validate results with your team before taking action.
Common Use Cases
Running a System-Wide Equity Audit
Before presenting to leadership or an equity task force, your team wants to understand the scope of language concerns across your full spec library. Run the bulk report to get a jurisdiction-wide snapshot: how many specs contain gender-specific terms, how many have accessibility concerns, and where the highest-priority reviews should go.
Prioritizing a Spec Modernization Project
Your agency is undertaking a classification modernization initiative. Instead of reviewing specs in alphabetical order, use the bulk report to triage, starting with specs flagged Review Needed and working down.
Cleaning Up Language During a Spec Rewrite
You're already rewriting a single spec, for a reclassification, a recruitment, or a routine update. Run the single-spec analysis before you finalize the draft so flagged language gets fixed in the same revision, instead of surfacing later in a bulk scan and requiring a second round of edits.
Running It on One Spec or Many
You can run this analysis two ways, depending on the question you're answering:
On a single spec, from the spec's Analyze menu, when you're updating one classification and want to clean up its language as part of the rewrite. See How to Run an Analysis on a Class Specification.
Across your catalog, from the Reports section, when you want a jurisdiction-wide picture, such as before an equity audit or a modernization project. See Running Reports for Bulk Analysis.
A common workflow combines the two: run the bulk report to find the specs that need attention, then run the single-spec analysis as you work through each one.
Note: Your agency's configuration determines which analyses are included for your account, so this analysis may not be available to you. Questions about your configuration? Contact Holly Support at [email protected].
Understanding Your Results
Single-Spec Results
When you run the analysis on an individual spec, results appear in the floating analysis window, organized by the four dimensions above. For each dimension, Holly identifies the specific language it flagged, explains the concern, and suggests replacement language you can apply as you edit the spec.
Work through the findings dimension by dimension. Flagged language is a starting point for your review, not a verdict: some flags will be genuine issues, and others will be justified by the role once you apply context the AI doesn't have.
Bulk Report Results
The bulk report shows one row per spec, with each spec assigned one of three statuses:
Review Needed: Specs with significant concerns. Start here.
Review Suggested: Specs that may benefit from updates. Hover over any flagged issue to see a plain-language explanation and a suggested fix.
Compliant: Specs where Holly found no issues across the four dimensions. Review these last, or skip them.
The hover explanations show the exact language that triggered each flag, so you can evaluate a finding without opening the spec. When you export the report to Excel, those explanations are included alongside every column in the table.
Filtering Your Results
Use the filter controls above the bulk results table to build a targeted review queue:
Filter by flag: Review Needed, Review Suggested, or Compliant.
Filter by issue type: Gender Language, Bias Language, Accessibility, or Fair Requirements.
Click any column header to sort. Combining a filter with a sort, for example filtering to Review Needed and sorting by Bias Language, is a fast way to decide where your team's review time goes first.
Custom Language Criteria
If your agency is working with an equity office to define specific language criteria, those custom keyword lists can be incorporated in a future update. Reach out to Holly Support at [email protected] to discuss your needs.