An Analysis of Negative Comments on Spanish Online Newspaper Editorials

This paper presents an empirical study of negative comments on editorials collected from major Spanish newspaper sites, El País, El Mundo, and ABC. We have evaluated online negative comments via sentiment analysis and detection of hate speech, using advanced analytical tools such as Pysentimiento and Perspective API. Our experimental results show that negative comments constituted about 76% across the three newspapers, with El Mundo showing the highest and El País the lowest negativity rates, likely reflecting their political orientations. Using the Perspective API, 13% of comments were notably harmful. The primary harmfulness detected was toxicity (61%), followed by insults and profanity. The follow-up close examination has identified that despite advancements, automated tools still struggle with context-specific negativity, such as sarcasm, highlighting the need for ongoing improvements in sentiment analysis and hate speech detection.