Linguistic Predictors of Rumor Veracity on the Internet

Abstract

This paper attempts to investigate the role of language in predicting the veracity of rumors on the Internet. Specifically, it seeks to examine rumor veracity as a function of six groups of linguistic predictors. These include rumors' (1) comprehensibility, (2) sentiment, (3) time-orientation, (4) quantitative details, (5) writing style, and (6) topic. A dataset of 2,391 rumors, about 20% of which were true and the rest false, drawn from the rumor-verification website Snopes.comwas used for investigation. The operationalized measures of the linguistic predictors were calculated for all rumors using the Linguistic Inquiry and Word Count (LIWC) tool. Binomial logistic regression was used for data analysis. The model performed generally well. The results specifically indicated that rumor veracity could be predicated by comprehensibility, time-orientation, writing style and topic of rumors.