St Reading Cross-genre Feature Comparisons for Spoken Sentence Segmentation 5

Automatic sentence segmentation of spoken language is an important precursor to downstream natural language processing. Previous studies combine lexical and prosodic fea19 tures, but can impose significant computational challenges because of the large size of feature sets. Little is understood about which features most benefit performance, partic21 ularly for speech data from different speaking styles. We compare sentence segmentation for speech from broadcast news versus natural multi-party meetings, using identical 23 lexical and prosodic feature sets across genres. Results based on boosting and forward selection for this task show that (1) features sets can be reduced with little or no loss in 25 performance, and (2) the contribution of different feature types differs significantly by genre. We conclude that more efficient approaches to sentence segmentation and similar 27 tasks can be achieved, especially if genre differences are taken into account.