Dedicated Language Resources for Interdisciplinary Research on Multiword Expressions: Best Thing since Sliced Bread

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 4418–4425 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 4418 Dedicated Language Resources for Interdisciplinary Research on Multiword Expressions: Best Thing since Sliced Bread Ferdy Hubers, Catia Cucchiarini, Helmer Strik Centre for Language Studies, Centre for Language and Speech Technology, Radboud University Erasmusplein 1, 6525 HT, Nijmegen, the Netherlands {f.hubers,c.cucchiarini,w.strik}@let.ru.nl Abstract Multiword expressions such as idioms (beat about the bush), collocations (plastic surgery) and lexical bundles (in the middle of) are challenging for disciplines like Natural Language Processing (NLP), psycholinguistics and second language acquisition, due to their more or less fixed character. Idiomatic expressions are especially problematic, because they convey a figurative meaning that cannot always be inferred from the literal meanings of the component words. Researchers acknowledge that important properties that characterize idioms such as frequency of exposure, familiarity, transparency, and imageability, should be taken into account in research, but these are typically properties that rely on subjective judgments. This is probably one of the reasons why many studies that investigated idiomatic expressions collected limited information about idiom properties for very small numbers of idioms only. In this paper we report on cross-boundary work aimed at developing a set of tools and language resources that are considered crucial for this kind of multifaceted research. We discuss the results of our research and suggest possible avenues for future research. Keywords: Multiword expressions, language resources, interdisciplinary research, idioms 1. Introduction Multiword expressions (MWEs), also known as formulaic expressions, are recurrent sequences of words that tend to co-occur in more or less fixed combinations. These MWEs come in different types, such as idiomatic expressions (hit the sack), collocations (make a decision, plastic surgery), binomials (back and forth), lexical bundles (in the middle of), and speech formulas (I beg your pardon) (Cacciari, 2014; Nattinger & DeCarrico, 1992; Siyanova-Chanturia & Martinez, 2014). MWEs in general, and idiomatic expressions in particular, are challenging from the perspective of Natural Language Processing (NLP), psycholinguistics and second language acquisition (SLA), and are therefore widely studied in these fields. From the perspective of NLP, MWEs are challenging because they pose serious difficulties in terms of processing in language and speech technology. Systems that use only compositional methods to generate text without taking into account the special characteristics of MWEs, can result in the generation of non-native-like constructions such as plastic operation. This is called the overgeneration problem (Sag et al., 2002). Expressions that convey a figurative meaning that cannot be extracted from the individual word meanings, can lead to parsing problems. These difficulties are often referred to as the idiomaticity problem (Sag et al., 2002). Both problems are challenging for applications that rely on semantic information, such as systems for automatic translation and automatic summarization. In order to overcome the overgeneration and idiomaticity problem, many researchers have been trying to automatically identify MWEs and more specifically idiomatic expressions (e.g., Constant et al., 2017; Cruys & Villada Moirón, 2007; Odijk, 2013; Ramisch, …

Content truncated to 500 words