Longitudinal analysis of sentiment and emotion in news media headlines using automated labelling with Transformer language models
Лонгитюдный анализ тональности и эмоций в заголовках новостных СМИ с использованием автоматической разметки языковыми моделями Transformer
2022-10-18
SCID: 54.1/sgf9b779
Discuss with AI
Transformer language modelsemotion detectionlongitudinal analysisnews media headlinessentiment analysis
Figures from the paper
Abstract (AI)
This work describes a chronological (2000-2019) analysis of sentiment and emotion in 23 million headlines from 47 news media outlets popular in the United States. We use Transformer language models fine-tuned for detection of sentiment (positive, negative) and Ekman's six basic emotions (anger, disgust, fear, joy, sadness, surprise) plus neutral to automatically label the headlines. Results show an increase of sentiment negativity in headlines across written news media since the year 2000. Headlines from right-leaning news media have been, on average, consistently more negative than headlines from left-leaning outlets over the entire studied time period. The chronological analysis of headlines emotionality shows a growing proportion of headlines denoting anger, fear, disgust and sadness and a decrease in the prevalence of emotionally neutral headlines across the studied outlets over the 2000-2019 interval. The prevalence of headlines denoting anger appears to be higher, on average, in right-leaning news outlets than in left-leaning news media.
Key Findings
1
Headline sentiment became increasingly negative across written news media between 2000 and 2019.
2
Headlines increasingly expressed anger, fear, disgust, and sadness, while emotionally neutral headlines became less prevalent; anger was more common in right-leaning outlets.
3
Right-leaning outlets had consistently more negative headlines than left-leaning outlets throughout the study period.
4
The study analyzes sentiment and emotion in 23 million headlines from 47 U.S.-popular news outlets spanning 2000–2019.
5
Transformer language models were fine-tuned to automatically label headline sentiment and Ekman’s six basic emotions, plus neutral.
Research Object
News media headlines from 47 U.S.-popular outlets published from 2000 to 2019
Research Subject
Longitudinal patterns and political-orientation differences in headline sentiment negativity and the prevalence of six basic emotions and neutrality
Publication Details
Publication Date
2022-10-18
Journal
Publisher
ISSN
Cited by
104
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai6
Exploiting Generative AI to Scale up Intelligent Tutoring Systems2023
Content Analysis: An Introduction to its Methodology.1984
The Content Analysis Guidebook2017
HuggingFace's Transformers: State-of-the-art Natural Language Processing2019
Out-group animosity drives engagement on social media2021
Qualitative Spatial Reasoning over Questions (Short Paper)2022