Multiword expressions in Natural Language Processing

The present volume addresses the way in which multiword expression resources are used in real-life downstream Natural Language Processing applications, how they interact with neural architectures and generative pre-trained language models and, more broadly, how linguistic theory and computational modeling can inform each other. It critically examines the properties of multiword expression resources and their integration into – or exclusion from – contemporary Natural Language Processing systems dominated by machine learning and Large Language Models. The contributions collected here investigate the morphology, syntax, and semantics of deadjectival verbs across the older Indo-European languages. By situating these formations within the broader typology of adjective derivation, property concept encoding, and primary vs. secondary verb formation, the volume forges new connections between typological generalizations, theoretical models, and historical-comparative evidence. In doing so, it not only illuminates the dynamics of verbal derivation in Indo-European but also advances our understanding of how the pathways from property concepts to verbal predicates are structured cross-linguistically.