The DELPH-IN Grammary: A Curated Repository of Grammars and Treebanks

Abstract

Precision computational grammars encode detailed linguistic analyses and compositional semantics that support rigorous investigation of grammatical phenomena, but their development requires substantial expertise and maintenance. To ensure long-term sustainability and accessibility of these resources, we present the DELPH-IN Grammary, a curated collection of twenty three HPSG grammars spanning seventeen languages and eight language families. The repository includes mature broad-coverage grammars (English, German, Japanese, Norwegian, Spanish) with associated treebanks, as well as grammars for typologically diverse and less-resourced languages. Each grammar is standardized with metadata, compiled using the ACE parser/generator, and loaded into the Linguistic Type Database for detailed inspection. Following FAIR principles, all resources are version-controlled and archived on Zenodo with annual releases synchronized to community development cycles. The Grammary enables reproducible grammar research, cross-linguistic typological studies, semantic parsing development, and grammar engineering pedagogy, providing the depth and theoretical grounding that complements data-driven approaches in computational linguistics. Our goal is to establish a sustainable model for preserving these valuable resources which bridge the critical gap between theoretical linguistics and empirical corpus based research. The Grammary is available at https://github.com/delph-in/grammary (Zenodo doi: zenodo.18945956).

Resources

Details