This lecture will explore how the advances in Natural Language Processing (NLP), Machine Learning and Artificial Intelligence have enabled the design of foundation language models that have successfully been incorporated into state-of-the-art in language technology tasks and applications, including machine translation, text simplification and dialogue systems; and therefore could provide an attractive alternative for accurately determining meaning in language. However, these models still face a serious challenge when dealing with non-literal language, like that involved in Multiword Expressions (MWEs) such as idioms (make ends meet), light verb constructions (give a sigh), verb particle constructions (shake up) and noun compounds (loan shark). These expressions are an integral part of the mental lexicon of native speakers often used to express complex ideas in a simple and conventionalised way accepted by a given linguistic community. Although they may display a wealth of idiosyncrasies, and may require knowledge that goes beyond what can be gathered from their individual words (e.g. “dark horse” not referring to an animal, but as an unknown candidate who unexpectedly succeeds), representing a real challenge for current NLP techniques, their accurate integration has the potential for improving the precision, naturalness and fluency of downstream tasks. In this talk, Aline will present an overview of how advances in foundational models have made an impact in the accurate processing of idiomaticity, concentrating on what models seem to incorporate of idiomatic interpretations, and where there is still work to be done.