Situational Irony in Farcical News Headlines

Abstract

Current approaches to the computational modelling of irony mostly address verbal irony and sarcasm, neglecting other productive types of irony, namely situational irony. The function of situational irony is to lay emphasis on (real or fictional) events that evoke peculiar and unexpected images, which usually create a comical effect on the audience. In this paper, we investigate the linguistic and rhetorical devices underlying this phenomenon in a corpus composed of farcical news headlines, aiming at its automatic recognition. Based on a thorough annotation study, we found that in news headlines unexpectedness is mainly achieved by combining terms from different conceptual domains (what we have called out-of-domain contrast). We then explored features for automatically identifying these semantic and pragmatic incongruities and evaluated their discriminating power in a corpus whose irony is expressed by means of out-of-domain contrast. The features explored in our experiments are globally effective in capturing this phenomenon, attaining a six percent improvement in terms of the F-Measure over a baseline that only considers lexical information. Moreover, we observed that the best features typically reported in the literature for identifying incongruity in sarcastic text are not relevant for detecting situational irony in farcical news, thus reinforcing the idea that these phenomena pose different challenges that require distinct modelling approaches.