From Incidents to Framing: A Dutch and English Frame-semantic Corpus and Lexicon
Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026)
Abstract
This paper reports on the final results of the Dutch FrameNet project. The project followed a new approach to aggregate an event corpus starting from registered events in Wikidata and collecting text in different languages that refer to these events. The resulting corpus is not only referentially grounded, but it is also grouped by the type of event, e.g. mass shootings, elections, sports events. A subset of the texts has been annotated with FrameNets frames for all references to the registered events and participants. The result is a unique corpus with comparable texts across languages that make reference to the same and similar events. From the annotations, we derived Dutch and English FrameNet lexicons, as well as reference lexicons. These lexicons allow us to infer abstractions from the annotations that also reflect sociocultural differences in framing the same entities and events.