Skip to main content

Open Dutch FrameNet

Welcome to the Open Dutch FrameNet! This page contains pointers to our lexicon and corpus:

Dutch FrameNet lexicon: https://github.com/cltl/OpenDutchFrameNetLexicon/tree/main

Dutch FrameNet data: https://github.com/cltl/OpenDutchFrameNetData

We also present an overview of the resources below. You can read a detailed description in the following publication:

Piek Vossen, Pia Sommerauer, Levi Remijnse. From incidents to framing: a Dutch and English frame-semantic corpus and lexicon. LREC 2026. [link to be added].

Data-to-text approach​

We collected texts covering different types of events to capture a large spectrum of variation in framing. We used Wikidata and other event registeries to collect texts written about specific events.

Lexicon overview​

The lexicon derived from the data-to-text corpus has a total of 6,964 entries. More detailed information is shown below:

Dutch FrameNet v1.0Counts
Total entries6,964
Nouns2,883
Names218
Verbs2,621
Adjectives530
Frames per entry3.01
Annotation per entry12.41
Total annotations28,670
Frame coverage2,311
Manual annotations27,841
Lexical baseline829
Data-to-text annotations22,435
RBN-Wordnet-FrameNet mappings954
Sonar annotations5,281

Corpus overview​

Using the data-to-text approach, we have collected reference texts in English and Dutch. A subset of the texts has been human-annotated for frames related to the incident of interest. Another subset of the English texts has been machine-labeled with OpenSesame (for all frames).

totalennlboth
types38372827
incidents22,05822,0022,3332,277
texts28,64225,6482,993n.a.
human-annotated1,312975337n.a.
system-annotated24,67024,6700n.a.
primary reference texts7,9036,7291,173n.a.
human-annotated1,312975337n.a.
system-annotated5,7515,7510n.a.
secondary reference texts20,73918,9191,820n.a.
human-annotated000n.a.
system-annotated18,91918,9190n.a.
word-count4,839,8444,431,000408,251n.a.
human-annotated699,876583,683116,193n.a.
system-annotated3,844,7783,844,7780n.a.