What can software tell us about media coverage and public opinion? An analysis of political news posts and audience comments on facebook by computerised method

Abstract

In this exploratory study, we applied an automated linguistic analysis method (TextMind) to a social movement context by comparing a sample set of online news posts (N1 = 13,434) with audience comments to the posts (N2 = 1,998,095) on Facebook. The findings of this study revealed that there were, in fact, linguistic differences between the news posts by news media outlets and their corresponding audience comments. TextMind is able to detect such linguistic differences and their changes over time. Comparative findings suggest: (1) The linguistic choices of news reporting are affected by news media’s (or journalists’) political, ideological, and market orientations. (2) The language used by traditional newspapers is not necessarily more conservative or moderate in emotion than their online competitors. (3) Linguistic choices in news posts would change over periods of time. However, (4) the language patterns of news posts did not directly affect linguistic choices of audiences in opinion expression, which remained relatively consistent.