The Patterns of Influence: LIWC Analysis of Leading News Portals' Impact and Communication Accommodation Theory on Twitter

Abstract

A quantitative analysis was conducted on the tweets of 20 leading Serbian mass media outlets and their followers during the three months from January 1 to March 30, 2021. The analysis, that encompassed more than 1 million tweets, was conducted using (a) Twitter API, (b) the Serbian version of LIWC linguistic dictionary, and (c) additional statistical analyses. Analyses revealed that words related to affective processes (negative emotions), social processes (family and friends), perceptual processes (feel and hear), and personal concerns (work and religion) expressed in tweets of mass media were predictors of the same words in not necessarily related tweets of their followers. The research has shown that spread of psychological and linguistic patterns by media and social networking sites is not limited to negative emotions solely. This research highlights the significance of recommender systems, as they help to determine the content that social media users encounter on a personal level. The findings also indicate how we can address emerging and pressing issues that affect individuals' well-being and democratic capacity, such as echo chambers and polarization. The obtained findings are in line with the communication accommodation theory, emotional contagion theory and negativity bias. Limitations of the present research, future directions, and implications are discussed.