What features characterize the writing style in the media coverage of the Graz case?
Based on the results of Corley and Wedeking (2014) who argue that certainty or “authoritativeness” is a basic but unrecognized tool to policy makers, we assumed that media reporting in the Graz case is more authoritative than in other neonaticide cases.
Based on Li et al. (2019), we assume that an analytic writing style has more power to influence the reader. Consequently, we think that media coverage of the Graz case—which is characterized by more insight words—inspires the reader to process information in a negative and humiliating way compared with media coverage of other neonaticide cases.
Regarding the influence of emotional language, Black et al. (2016) could show that judges are more likely to vote for parties whose briefs are written in an objective language with few emotion words. Briefs using emotional language are less likely to persuade the Supreme Court. We conclude that emotion words can decrease credibility in a context where official language is needed.
According to many authors like Pressman and Cohen (2007), social words are an indirect measurement of social relationships, and in Ritter’s study, Christians mention social processes more often suggesting stronger relationships and support networks and more positive and fewer negative emotions (Ritter et al. 2014). Because of the extensive media coverage in a negative and humiliating way, we assume that the perpetrator is not described within her social network but is described as isolated and in a state of emergency.
Included media reports
We analysed Austrian print media reports published between December 16, 2004, and December 16, 2014 from the Austrian Press Agency that included the word “Babymord” (killing of a baby), “Kindesmord” (neonaticide, child murder), or “Kindestötung” (infanticide). The majority of the newspapers were daily newspapers. All national newspapers (e.g., Kronen Zeitung, Kurier) as well as those regional newspapers with the highest circulation in Austria (e.g., Kleine Zeitung, Tiroler Tageszeitung) were included in the analysis. The target audience of these newspapers was highly diverse in terms of socio-demographics and political orientation. Our search yielded 1027 articles. We excluded 696 articles, because the focus was on another topic: 228 focused on fictional stories, 11 articles described war victims and their children, one article mentioned child murder as opposed to murder of an adult, 85 articles were duplicates, 81 articles were letters from readers, 75 articles were texts from Austrian press agency (not newspaper articles), nine articles focused on murdering of pets, and 206 articles focused on physical abuse of children. In total, 331 articles fulfilled the inclusion criteria and were included in the final data analysis (see Table 1).
The Linguistic Inquiry and Word Count (LIWC) software (Pennebaker et al. 2001) was used to analyse the wording of the texts. LIWC is a word count-based text analysis program, which uses distinct word categories with more than 2300 words or word stems. LIWC operates by comparing all words in a given text to a dictionary and by counting the number of words in this text that fall into specific word categories relative to the total number of words (text length), i.e., the percentage of all of the words in that text (Pennebaker et al. 2001). We used the German translation of LIWC by Wolf et al. (2008). The German LIWC categories have been shown to have high equivalence to their English counterparts based on two studies using 122 bilingual text units and 104 emails.
The goal of LIWC is to use objective linguistic data to gain information about an individual’s cognitive processing, including attentional focus, emotionality and thinking styles. Exploring media reports with LIWC means shedding light on cultural scripts of a society. In our study, our basic research question is whether there are differences between the language of the Graz case and other cases of neonaticides in Austrian newspapers. Regarding the articles, there are two different categories of text differing in writing style: articles on the case itself on one hand and articles about the trial/prosecution of the perpetrators on the other hand. Therefor the scores for all LIWC categories were subjected to a case (case in Graz vs. other cases) × coverage (case coverage vs. trial coverage) analysis using generalized linear models. Because our residuals were not normally distributed, we conducted generalized linear models with gamma distributions (Venables and Ripley 2002).