Grammatical Factor and Spoken Sentence Recognition

Experiments point at the need for a robust grammatical analysis on recognition results. A robust parsing can rst, help distinguish the incorrect recognitions that are recoverable; second, produce a relevant analysis for the interpretation process. We review existing robust parsing techniques and propose an approach based on Lexicalized Tree Grammars. We report experiments conducted on \Virtual Speaker", an application devel-opped at Thomson-CSF Corporate Research Lab., which aims at helping users chose their TV-program through a dialogue interface. We tested diierent speakers and two benchs of approximatly 50 sentences each, on three SR (Speech Recognition) systems 1 presented in Table 1, where CFG stands for Context Free Grammar and LM stands for Language Model. In Figure 1 the CFG-driven system performs better than LM-driven systems, since almost all utterances tested are in the CFG grammar. However, performance patterns have common features: Correct, slighlty incorrect and totally