Five-category feature taxonomy for classifying ethnic bias in educational texts¶

Theory · All theories
Evidence · 2 claims (2 for) · 1 study (1 associational), q2 · 1 of 1 report an effect size · 2 claims rest on one study

Description¶

The article organizes candidate bias signals into a taxonomy of "five categories: psycholinguistic markers, linguistic abstraction, language valence, ethnic group mentions, and contextual attributes". Psycholinguistic markers (e.g., aggressiveness, emotional stability, self-reference) reflect the author's psychological state; abstraction metrics capture specificity versus generality in descriptions; valence quantifies emotional tone; mentions operationalize ethnic group references via a keyword list of over 1,000 terms; contextual attributes encode discipline and material type. Features were computed for both sentences and surrounding excerpts to make classification more context aware.

Design Implications¶

Context¶

Requirements¶

• Part-of-speech tagging, psycholinguistic ratio computation, and sentiment (VADER) scoring pipelines over sentences and excerpts
• OER metadata for contextual attributes and a keyword list for ethnic group mentions

Constraints¶

• Group references alone do not necessarily indicate bias and are more informative alongside attributes such as valence

Target Learners¶

• Higher education students
• Online learners

Target Learning Objectives¶

• Interpretable identification of ethnic bias in text-based learning materials

Claims¶

• Linguistic Features Correlate Ethnic Bias [+M]
• Social sciences content is positively correlated with perceived ethnic bias in text-based learning materials [+W]

Related Theories¶

Examples¶

-

Key Sources¶

• Lb0