Flexible Corpus Annotation with Property Grammars

International audienceWe present in this paper a syntactic annotation project relying on a linguistic formalism called Property Grammars. This constraint-based formalism allows to encode different levels of syntactic information (the granularity description can be tuned according to the set of constraints to be verified). Moreover, this approach allows to propose a description whatever the form of the input (being it grammatical or not). After a presentation of the formalism and its representation, we describe its use in the annotation of a spoken language corpus