Neutral or Framed? A Sentiment Analysis of 2019 Abortion Laws

Area of Science:

• Computational Linguistics
• Legal Studies
• Political Science

Background:

• Abortion law remains a contentious issue, with legislative language potentially influencing judicial interpretation.
• Sentiment analysis (SA), a Natural Language Processing (NLP) technique, can detect emotive language and identify framing in text.
• Understanding the semantic structure of legal documents is crucial for assessing potential biases.

Purpose of the Study:

• To apply sentiment analysis (SA) to evaluate the semantic framing of restrictive and protective abortion bills enacted in 2019.
• To determine if SA can identify intentional wording biases in legal texts.
• To explore the implications of language framing on the interpretation of abortion laws.

Main Methods:

• A sample of 19 abortion bills from 2019 was identified using Legiscan.
• Bills were categorized as restrictive (n=12) or protective (n=7).
• Aggregate and separate sentiment analysis was performed, interpreting scores qualitatively to assess language positivity/negativity.

Main Results:

• Overall, 56% of the text in the analyzed bills comprised negative terms.
• Restrictive bills exhibited more negative language (67%) compared to protective bills (58%).
• Sentiment scores varied widely, with some bills showing neutrality (score of 0), contrasting with the US Constitution's score of 1.

Conclusions:

• Sentiment analysis is a viable method for detecting language biases in legal documents.
• The analyzed abortion bills demonstrated framing aligned with political ideologies.
• The study provides evidence that neutral language in legislation is attainable.