Figurative messages and affect in Twitter: Differences between #irony, #sarcasm and #not

Abstract

The use of irony and sarcasm has been proven to be a pervasive phenomenon in social media posing a challenge to sentiment analysis systems. Such devices, in fact, can influence and twist the polarity of an utterance in different ways. A new dataset of over 10,000 tweets including a high variety of figurative language types, manually annotated with sentiment scores, has been released in the context of the task 11 of SemEval-2015.In this paper, we propose an analysis of the tweets in the dataset to investigate the open research issue of how separated figurative linguistic phenomena irony and sarcasm are, with a special focus on the role of features related to the multi-faceted affective information expressed in such texts. We considered for our analysis tweets tagged with #irony and #sarcasm, and also the tag #not, which has not been studied in depth before. A distribution and correlation analysis over a set of features, including a wide variety of psycholinguistic and emotional features, suggests arguments for the separation between irony and sarcasm. The outcome is a novel set of sentiment, structural and psycholinguistic features evaluated in binary classification experiments. We report about classification experiments carried out on a previously used corpus for #irony vs #sarcasm. We outperform in terms of F-measure the state-of-the-art results on this dataset. Overall, our results confirm the difficulty of the task, but introduce new datadriven arguments for the separation between #irony and #sarcasm. Interestingly, #not 1 Corresponding author: sulis@di.unito.it. 2 The first two authors equally contributed to this work.

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

“…This issue is challenging, and only recently addressed from computational linguistics. In particular, new data-driven arguments for a possible separation between irony and sarcasm emerged from recent work on Twitter data (Sulis et al, 2016). It could be interesting to see the relation between the finer-grained and pragmatic phenomena related to irony investigated in the present study and the higher-level distinction between irony and sarcasm.…”

The Impact of Figurative Language on Sentiment Analysis

“…Sulis et al (2016) analyse the corpus from Semeval-2015 Task 11 in terms of hashtags (#irony, #sarcasm, and #not) and confirm that messages using figurative language mostly express a negative sentiment. They experimented with 1 They used ensemble methods and ridge regression.…”

“…Although these works succeeded in their goal to detect variations of sarcasm, one final step is still missing -the evaluation of sentiment analysis with and without additional sarcasm indicators. There have been attempts at investigating the impact of sarcasm on sentiment analysis (Maynard and Greenwood, 2014) or thorough analysis of hashtags indicating sarcastic tweets (Sulis et al, 2016). However, the impact of figurative language (including sarcasm) on sentiment analysis has not yet been studied in depth.…”

Proceedings of the First Workshop on Abusive Language Online

“…Existing work identifying sarcasm on Twitter (Sulis et al, 2016;Ling and Klinger, 2016;Wang, 2013) finds that sarcastic tweets tend to express pejorative meaning with positive words. The sarcastic instances in our data show a different pattern, using pejorative nominalizations with other negative words to mock discriminatory mindsets, in the end conveying negative sentiment towards those who use this type of abusive language.…”