What's in a Line? Using Natural Language Processing on Teleplay Dialogue to Evaluate On-Screen Representation
Year: 2025
Type: article
Abstract: Research has identified many social benefits of positive on-screen representation, such as improved self-esteem and social inclusion, as well as the harms that poor representation, such as stereotyping, can produce. While many efforts exist to promote these benefits and reduce these harms, much evaluation of this work focuses only on representation quantity, leaving a critical component in understanding the impacts of these efforts unexamined. To evaluate on-screen representation quality, I propose a framework that uses dialogue in television scripts to quantify what the Stanford SPARQ Lab defines as the ``3 P''s (presence, positivity, and positionality), informed in part by measures specified in the empirically tested Stereotype Content Model. My analysis of presence explores several network and quantitative metrics using television dialogue, while I employ state-of-the-art large language models with human annotation data in an extensive analysis of positivity and positionality in teleplays. Overall, these experiments reveal the potential of this framework to identify informative, meaningful patterns and use various computational and natural language processing techniques in crafting concrete metrics for the assessment and advancement of on-screen representation.
Source: Stanford Digital Repository
DOI: 10.25740/bd273jz3725
Author: Eigbe, Naomi
Language: English
Subfield: General Social Sciences
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W6908178672
FWCI:
Cites:
Cited by:
Related to: