A new approach for measuring subjectivity in Brazilian news

Abstract

With the advent of digital journalism, information democratization has become a reality since news articles are published as soon as the facts occur, and that they are accessible from any device connected to the internet. It is common sense the perception that some media outlets are more biased than others when it comes to the way of exposing the facts. However, automatic ways of measuring such biases is still an open research challenge. Under the assumption that journalistic texts must have an objective and impartial language, high levels of subjectivity in these texts may indicate bias. This paper proposes an initial analysis on the usage of subjectivity lexicons to characterize subjectivity in seven popular media outlets in Brazil. To better understand the obtained results, we carried out a correlation analysis between the levels of subjectivity, readability, and news popularity metrics. The adopted methods, along with the findings obtained from this research, may contribute to a better understanding of the linguistic characteristics of the news that readers consume daily in Brazil.