Without lexicons, multiword expression identification will never fly: A position statement

Abstract

Because most multiword expressions (MWEs), especially verbal ones, are semantically non-compositional, their automatic identification in running text is a prerequisite for semantically-oriented downstream applications. However, recent developments, driven notably by the PARSEME shared task on automatic identification of verbal MWEs, show that this task is harder than related tasks, despite recent contributions both in multilingual corpus annotation and in computational models. In this paper, we analyse possible reasons for this state of affairs. They lie in the nature of the MWE phenomenon, as well as in its distributional properties. We also offer a comparative analysis of the state-of-the-art systems, which exhibit particularly strong sensitivity to unseen data. On this basis, we claim that, in order to make strong headway in MWE identification, the community should bend its mind into coupling identification of MWEs with their discovery, via syntactic MWE lexicons. Such lexicons need not necessarily achieve a linguistically complete modelling of MWEs' behavior, but they should provide minimal morphosyntactic information to cover some potential uses, so as to complement existing MWE-annotated corpora. We define requirements for such a minimal NLP-oriented lexicon, and we propose a roadmap for the MWE community driven by these requirements.

Verbal Multiword Expression Identification: Do We Need a Sledgehammer to Crack a Nut?

“…2) achieve global cross-language macro-averaged F1 scores below 0.6. Savary et al (2019a) argue that this is mainly due to the very nature of MWEs. Namely, MWEs of the general language (as opposed to specialized phenomena such as named entities and multiword terms) are mostly regular at the level of tokens (individual occurrences), but idiosyncratic at the level of types (sets of occurrences).…”

“…This implies strong lexicalization, that is, it is the combination of precise lexemes (and not so much of their senses) which makes a MWE. Savary et al (2019a) claim that, due to these properties, the generalization power of mainstream machine learning is relatively weak for MWE identification. This fact is confirmed by the results in the present paper, in which we outperform learning-based state-of-the-art systems using simple and interpretable rules and filters.…”

“…This shows that thorough account of seen VMWEs greatly boosts global MWE identification. In other words, with competing systems being extremely limited to deal with unseen VMWEs (Savary et al, 2019a), our method's superiority on (more frequent) seen VMWEs compensates the absence of treatment for unseen ones.…”

Pre-tokenization of Multi-word Expressions in Cross-lingual Word Embeddings

“…However, such training resources are available only in a limited number of languages, and even with such resources, the automatic analysis of MWEs is known to be very difficult. Savary et al (2019) argues the importance of syntactic MWE lexicons for further development in this area.…”

“…1 In this study, we employ a simple method to identify MWEs in corpora by using MWE dictionaries instead of automatic detection. Despite the rich body of work (Constant et al, 2017), including methods developed in specialized shared tasks (Schneider et al, 2014;Savary et al, 2017;Ramisch et al, 2018), automatic MWE detection is still a hard problem (Savary et al, 2019). Ramisch et al (2012) tested several unsupervised discovery methods and reported that they performed poorly in terms of either precision or recall.…”