Exploring the Impact of Pragmatic Phenomena on Irony Detection in Tweets: A Multilingual Corpus Study

Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pages 262–272, Valencia, Spain, April 3-7, 2017. c⃝2017 Association for Computational Linguistics Exploring the Impact of Pragmatic Phenomena on Irony Detection in Tweets: A Multilingual Corpus Study Jihen Karoui1, Farah Benamara1, V´eronique Moriceau2, Viviana Patti3, Cristina Bosco3, and Nathalie Aussenac-Gilles1 1IRIT, CNRS, Universit´e de Toulouse, France 2LIMSI, CNRS, Univ. Paris-Sud, Universit´e Paris-Saclay, France 3Dipartimento di Informatica, University of Turin, Italy 1{karoui,benamara,aussenac}@irit.fr 2{moriceau}@limsi.fr 3{patti,bosco}@di.unito.it Abstract This paper provides a linguistic and prag- matic analysis of the phenomenon of irony in order to represent how Twitter’s users exploit irony devices within their commu- nication strategies for generating textual contents. We aim to measure the impact of a wide-range of pragmatic phenomena in the interpretation of irony, and to inves- tigate how these phenomena interact with contexts local to the tweet. Informed by linguistic theories, we propose for the first time a multi-layered annotation schema for irony and its application to a corpus of French, English and Italian tweets.We de- tail each layer, explore their interactions, and discuss our results according to a qual- itative and quantitative perspective. 1 Introduction Irony is a complex linguistic phenomenon widely studied in philosophy and linguistics (Grice et al., 1975; Sperber and Wilson, 1981; Utsumi, 1996). Glossing over differences across ap- proaches, irony can be defined as an incongruity between the literal meaning of an utterance and its intended meaning. For many researchers, irony overlaps with a variety of other figurative devices such as satire, parody, and sarcasm (Clark and Gerrig, 1984; Gibbs, 2000). In this paper, we use irony as an umbrella term that includes sarcasm, although some researchers make a distinction be- tween them, considering that sarcasm tends to be more aggressive (Lee and Katz, 1998; Clift, 1999). Different categories of irony have been studied in the linguistic literature such as hyperbole, exag- geration, repetition or change of register (see sec- tion 3 for a detailed description). These categories were mainly identified in literary texts (books, po- ems, etc.), and as far as we know, no one explored them in the context of social media. The goal of the paper is thus four folds: (1) analyse if these categories are also valid in social media contents, focusing on tweets which are short messages (140 characters) where the context may not be explic- itly represented; (2) examine whether these cate- gories are linguistically marked; (3) test if there is a correlation between the categories and mark- ers; and finally (4) see if different languages have a preference for different categories. This analysis can be exploited in a purpose of automatic irony detection, which is progres- sively gaining relevance within sentiment analy- sis (Maynard and Greenwood, 2014; Ghosh et al., 2015). In particular, it will bring out the most discriminant pragmatic features that need to be taken into account for an accurate irony detection, therefore helping systems improve beyond stan- dard approaches that still heavily rely on features …

Content truncated to 500 words