The aim of this project is to develop a Swedish so-called constructicon, a database of Swedish constructions. It is meant to be a large-scale electronic resource for linguistic, lexicographic, and educational purposes, as well as for language technology applications. A special priority is to account for constructions that concern second language acquisition. We also mean to develop tools for automatic identification of constructions.
A language consists to quite a large extent of restricted, semi-productive constructions. Such structures tend to be highly problematic for language technology as well as for language learners, and also seem to be quite common. Yet, they are often neglected in both grammars and dictionaries, which focus on general rules and individual words, respectively. Hence, a comprehensive account of construction-specific patterns is so far lacking. We therefore intend to develop such a resource, aiming to make it descriptively adequate, simple enough for large-scale coverage, and formalized to enable computational uses.
The constructicon is developed in close collaboration with the Swedish FrameNet. The approach combines construction grammar, lexicography, and language technology, in a manner similar to the Berkeley FrameNet Constructicon, but it also builds on the Swedish lexicographic tradition, and has a particular L2 focus.