Polysemy in Spoken Conversations and Written Texts

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 1680–1690 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 1680 Polysemy in Spoken Conversations and Written Texts Aina Gar´ı Soler, Matthieu Labeau, Chlo´e Clavel LTCI, T´el´ecom-Paris, Institut Polytechnique de Paris, France {aina.garisoler,matthieu.labeau,chloe.clavel}@telecom-paris.fr Abstract Our discourses are full of potential lexical ambiguities, due in part to the pervasive use of words having multiple senses. Sometimes, one word may even be used in more than one sense throughout a text. But, to what extent is this true for different kinds of texts? Does the use of polysemous words change when a discourse involves two people, or when speakers have time to plan what to say? We investigate these questions by comparing the polysemy level of texts of different nature, with a focus on spontaneous spoken dialogs; unlike previous work which examines solely scripted, written, monolog-like data. We compare multiple metrics that presuppose different conceptualizations of text polysemy, i.e., they consider the observed or the potential number of senses of words, or their sense distribution in a discourse. We show that the polysemy level of texts varies greatly depending on the kind of text considered, with dialog and spoken discourses having generally a higher polysemy level than written monologs. Additionally, our results emphasize the need for relaxing the popular “one sense per discourse” hypothesis. Keywords: Semantics, Word Sense Disambiguation, Document classification / Text categorisation 1. Introduction Polysemous words are not the majority in the vocabu- lary of a language, but they are used frequently in all types of texts. These words may even be employed in their different senses within the same discourse without necessarily hindering communication. For example, speakers sometimes exploit polysemy to keep a con- versation interesting, with irony, jokes, or metaphors (Nerlich and Clarke, 2001). Gale et al. (1992) investigate whether multiple in- stances of a polysemous word in a discourse tend to share the same sense, and conclude that this is in- deed the case. They propose the “one sense per dis- course” (OSD) hypothesis, which, despite its popu- larity and convenience to be used as a heuristic for Word Sense Disambiguation (McCarthy and Carroll, 2003; Preiss and Stevenson, 2013; Pilehvar and Nav- igli, 2015; Chaplot and Salakhutdinov, 2018), has also been put into question (Krovetz, 1998). Works exploring the use of polysemous words in texts (Magnini et al., 2002; McCarthy et al., 2007) have fo- cused on a specific kind of discourse, namely scripted written monologs. In this paper, we want to compare the polysemy level of different kinds of texts. We hy- pothesize that the use of polysemous words in multiple senses within a discourse varies depending on the na- ture of the text. In order to verify this hypothesis, we compare the polysemy level of texts along three differ- ent axes: spoken/written, monolog/(dyadic) dialog, and spontaneous/scripted. In our analysis, the most promi- nent type of discourse is spoken language transcripts, and most of these consist of dialogs. There …

Content truncated to 500 words