Formal and Computational Linguistics (ComForT), Leuven
The Research Group conducts theoretical and descriptive research in the fields of formal and computational linguistics, a.o. with a view to its application in language and speech technology. The central aim is to acquire insight into the complexity of linguistic knowledge by studying the formal properties of the structures and rules that constitute the architecture of language and grammar.
The formal modelling of this knowledge is mainly carried out by means of structural and/or logical-mathematical representations. The group strives towards interdisciplinary collaboration with researchers from computer science/AI and analytical philosophy/logic.
Three tracks are distinguished within formal syntax:
- In the monostratal framework of Head-Driven Phrase structure Grammar the group studies predicative complements, object raising and verbal clusters, copular verbs and PPs a.o..
- In the generative framework of Minimalism the group studies ellipsis, expletives, complementizers, binding, control, particle verbs and the morphology-syntax interface.
- In the descriptive-structural framework the group studies the synchronic and diachronic morpho-syntax of Dutch and German, the structure of the NP, conditional clauses and connectives (adverbs and conjunctions).
Two tracks are distinguished within formal semantics:
- On the descriptive level, research concerns quantification in natural language (comparative and proportional quantification, negation and polarity, scope-relations) on the one hand, and tense, aspect, telicity and boundedness on the other.
- On the theoretical level, research concerns the logical geometry of relations of opposition, implication and duality on the one hand, and the algebra of colours, numbers and geometry on the other.
The group develops language technology apploications in the field of:
- automatic tagging, lemmatizing and parsing for Dutch
- hybrid systems for automatic translation from and into Dutch
- automatic text alignment
The group plays an active role in the creation of large-scale annotated corpora and tree banks for Dutch (D-coi, LASSY, SoNaR) and the development of NLP Tools and Resources in the broad academic framework of Digital Humanities.