PARSEME multilingual corpus of verbal multiword expressions


Multiword expressions (MWEs) are known as a ‘pain in the neck’ due to their idiosyncratic behaviour. While some categories of MWEs have been largely studied, verbal MWEs (VMWEs) such as to take a walk, to break one’s heart or to turn off have been relatively rarely modelled. We describe an initiative meant to bring about substantial progress in understanding, modelling and processing VMWEs. In this joint effort carried out within a European research network we elaborated a universal terminology and annotation methodology for VMWEs. Its main outcomes, available under open licenses, are unified annotation guidelines, and a corpus of over 5.4 million words and 62 thousand annotated VMWEs in 18 languages.

Multiword expressions at length and in depth: Extended papers from the MWE 2017 workshop,