Testing an Inductive Mixed-method Computational Approach to News Frame Analysis: An analysis of Hungarian online reporting of the 2014 Russia-Ukraine conflict

Abstract

Computational frame analysis is a highly contested, yet widely researched area of study, promising novel approaches for developing a broader understanding of news-reporting practices. Such an approach developed by Walter and Ophir (2019) is applied to inductively discover news frames used by Hungarian online news portals during the coverage of the events in Ukraine in 2014. The approach, named Analysis of Topic Model Networks, utilises LDA topic modelling and network community detection methods for inductively identifying frame packages. The applicability of the approach to Hungarian language text is explored using the BERTopic algorithm in place of LDA. Two subcorpora are analysed from the webarchive developed by Indig et al. (2019).