The Distribution of Affective Words in a Corpus of Newspaper Articles

This paper explores the possibility of automatically measuring and comparing affectiveness features in texts from different domains and sub-domains. More specifically our main research question concerns the distribution of affective words within the various subgenres of a newspaper corpus along three affective parameters: Valence (V), pleasantness, Arousal (A), the intensity of the emotion, and Dominance (D), the degree of control exerted by the perceiver over the stimulus. The study is largely based on work by Warriner et al. (in press), who recently divulged a study reporting affective ratings for approximately 14,000 English lemmas in terms of V, A and D. 100,000 token samples of newspaper language from 10 subsections of the Guardian newspaper were analyzed for the presence of these lemmas and the average V, A and D values for each subsection were calculated. Crime and Travel were seen to be those with the most atypical values.