Exact Grammaticality of Likely Parse Tree (Jugement exact de grammaticalité d'arbre syntaxique probable) [in French]

The robustness of probabilistic parsing generally comes at the expense of grammaticality judgment – the grammaticality of the most probable output parse remaining unknown. Parsers, such as the Stanford or the Reranking ones, can not discriminate between grammatical and ungrammatical probable parses, whether their surface realisations are themselves grammatical or not. In this paper we show that a Model-Theoretic representation of Syntax alleviates the grammaticality judgment on a parse tree. In order to demonstrate the practicality and usefulness of an alliance between stochastic parsing and knowledge-based representation, we introduce an exact method for putting a binary grammatical judgment on a probable phrase structure. We experiment with parse trees generated by a probabilistic parser. We show experimental evidence on parse trees generated by a probabilistic parser to confirm our hypothesis. Mots-clés : Jugement de grammaticalité, syntaxe par modèles, Grammaires de Propriétés, analyse syntaxique probabiliste.