Sentiment Analysis of Homeric Text: The 1st Book of Iliad

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 7071–7077 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 7071 Sentiment Analysis of Homeric Text: The 1st Book of Iliad John Pavlopoulos, Alexandros Xenos, Davide Picca Athens University of Economics and Business, Greece University of Lausanne, Switzerland {annis, a.xenos20}@aueb.gr, davide.picca@unil.ch Abstract Sentiment analysis studies are focused more on online customer reviews or social media, and less on literary studies. The problem is greater for ancient languages, where the linguistic expression of sentiments may diverge from modern linguistic forms. This work presents the outcome of a sentiment annotation task of the first Book of Iliad, an ancient Greek poem. The annotators were provided with verses translated into modern Greek and they annotated the perceived emotions and sentiments verse by verse. By estimating the fraction of annotators that found a verse as belonging to a specific sentiment class, we model the poem’s perceived sentiment as a multi-variate time series. By experimenting with a state of the art deep learning masked language model, pre-trained on modern Greek and fine-tuned to estimate the sentiment of our data, we registered a mean squared error of 0.063. This low error indicates that sentiment estimators built on our dataset can potentially be used as mechanical annotators, hence facilitating the ‘distant reading’ of Homeric text. Our dataset is released for public use. Keywords: Sentiment analysis, natural language processing, language resources, Homeric text 1. Introduction Sentiment analysis is a series of methods, techniques, and tools about detecting and extracting subjective in- formation, such as opinion and attitudes, from language (Liu, 2009). The field has experienced a rapid growth during the last years, with most of the thousands of pa- pers being published after 2004 (M¨antyl¨a et al., 2018). The main focus of sentiment analysis is the analysis of online customer reviews and, more recently, social media texts, but applications also range to financial market prediction, business review analysis, politics, demonetisation, crime prediction, and disaster assess- ment (Yadav and Vishwakarma, 2020). The challenge of sentiment analysis is more demand- ing for ancient sources for which the linguistic expres- sion of sentiments may be significantly divergent from modern linguistic forms. Nonetheless, attempting to overcome such an issue, researchers employed the cor- responding modern Chinese translation (Zhao et al., 2014) in an approach to classify the sentiment of an- cient Chinese literature. The same approach was fol- lowed by Yeruva et al. (2020), who used an English translation of randomly chosen sentences from Aeschy- lus’s ancient Greek tragedies to classify the sentiment. The authors found that in-context annotation (preced- ing verses) leads to different perceived sentiment com- pared to out-of-context annotation, which is probably due to the temporal order of sentiments in long texts. Motivated by the limited sentiment annotations for an- cient languages, specifically for ancient Greek, in this work we annotate an ancient Greek poem, translated in modern Greek, verse by verse, in order to maximise the context provided to the annotators. …

Content truncated to 500 words