Interactive Word Sense Disambiguation in Foreign Language Learning

Abstract

“Word sense awareness” is a feature which is not yet implemented in most corpus query tools, Intelligent Computer-Assisted Language Learning (ICALL) environments or computer-readable didactic resources such as graded word lists (Alfter and Graën, 2019; Pilán et al., 2016; Tack et al., 2018). The present paper aims to contribute to filling this lacuna by presenting a word sense disambiguation (WSD) method for ICALL purposes. The method, which is targeted at Spanish as a foreign language (SFL), takes a few prototypical example sentences as input, converts these sentences into “sense vectors”, and integrates part of the training data collection process into interactive vocabulary exercises. The evaluation of the method is based on a selection of 50 ambiguous items related to the domain of economics and compares different types of input data. With a top weighted F1 score of 0.8836, the present study shows that the currently available NLP tools, resources and methods provide all the necessary building blocks for developing a WSD method which can be integrated into interactive ICALL environments.

Making ICALL’s intelligence accessible and understandable: evaluation, validation, and future directions of ICALL ecosystems

“…Nevertheless, using ICALL also comes with its limitations. Recognising lexically ambiguous items and distinguishing between their senses is one of those pending issues in ICALL research, especially for languages other than English (Degraeuwe & Goethals, 2022b). In fact, the Natural Language Processing (NLP) driven technique of Word Sense Disambiguation (WSD) is rarely integrated in corpus query tools or in the development of computer readable resources for didactic purposes, such as graded word lists (Tack, 2021).…”