Longitudinal analysis of sentiment and emotion in news media headlines using automated labelling with Transformer language models - PubMed (original) (raw)

Longitudinal analysis of sentiment and emotion in news media headlines using automated labelling with Transformer language models

David Rozado et al. PLoS One. 2022.

Abstract

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.

PubMed Disclaimer

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Fig 1

Fig 1. The solid blue line shows the average yearly sentiment of headlines across 47 popular news media outlets.

The shaded area indicates the 95% confidence interval around the mean. A statistical test for the null hypothesis of zero slope is shown on the bottom left of the plot. The percentage change on average yearly sentiment across outlets between 2000 and 2019 is shown on the top left of the plot.

Fig 2

Fig 2. Average yearly sentiment of headlines grouped by the ideological leanings of news outlets using human ratings of outlets political bias from the 2019 AllSides Media Bias Chart v1.1 [24].

The figure displays the standard error bars of the average yearly sentiment for outlets within each color-coded political orientation category. For each ideological grouping, statistical tests for the null hypothesis of zero slope are shown on the bottom left of the plot.

Fig 3

Fig 3. Average yearly prevalence of news articles headlines denoting different types of emotionality in 47 popular news media outlets.

The shaded gray area indicates the 95% confidence interval around the mean. Note the different scale of the Y axes for the different emotion types. For each emotional category, statistical tests for the null hypothesis of zero slope are shown on the bottom left of each subplot. Reported p-values have been Bonferroni-corrected for multiple comparisons. The percentage changes between 2000 and 2019 are shown on the top left of each subplot.

Fig 4

Fig 4. Yearly prevalence of headlines denoting different types of emotionality in 47 popular news outlets grouped by human ratings of news media ideological leanings from the 2019 AllSides Media Bias Chart v1.1 [24].

Note the different scale of the Y axes for the different emotion types. Only statistical tests within each ideological grouping for which the null hypothesis of zero slope was rejected (after Bonferroni correction for multiple comparisons) are shown on the bottom left of each plot.

Similar articles

Cited by

References

    1. O’Brien H. L., “Exploring user engagement in online news interactions,” Proceedings of the American Society for Information Science and Technology, vol. 48, no. 1, pp. 1–10, 2011, 10.1002/meet.2011.14504801088. - DOI
    1. Ecker U. K. H., Lewandowsky S., Chang E. P., and Pillai R., “The effects of subtle misinformation in news headlines,” Journal of Experimental Psychology: Applied, vol. 20, no. 4, pp. 323–335, 2014, doi: 10.1037/xap0000028 - DOI - PubMed
    1. Bransford J. D. and Johnson M. K., “Contextual prerequisites for understanding: Some investigations of comprehension and recall,” Journal of Verbal Learning and Verbal Behavior, vol. 11, no. 6, pp. 717–726, Dec. 1972, doi: 10.1016/S0022-5371(72)80006-9 - DOI
    1. Hasell A., “Shared Emotion: The Social Amplification of Partisan News on Twitter,” Digital Journalism, vol. 9, no. 8, pp. 1085–1102, Sep. 2021, doi: 10.1080/21670811.2020.1831937 - DOI
    1. Berger J. and Milkman K. L., “What Makes Online Content Viral?,” Journal of Marketing Research, vol. 49, no. 2, pp. 192–205, Apr. 2012, doi: 10.1509/jmr.10.0353 - DOI

MeSH terms

Grants and funding

The author(s) received no specific funding for this work.

LinkOut - more resources