Corpus analysis reveals diverse pragmatic dimensions of directives in conversational English, highlighting benefits for natural language processing models.
This study introduces dassBNC2014, a manually annotated subset of the Spoken British National Corpus 2014 designed to capture directive speech acts in everyday conversation. In natural language processing (NLP), speech act annotations in open-domain dialog remain scarce, while manually annotated data are fragmented and inconsistently accessible. With dassBNC2014, we aim to address the persistent gaps in existing resources. Our annotation scheme combines the ISO 24617-2 framework with insights from speech act theory to achieve an intermediate level of granularity, which is both theoretically grounded and operationally feasible. Analysis of the annotated directives shows that they constitute a heterogeneous set of practices varying systematically in interactional locus, temporal orientation, deontic strength and action type. These findings refine the theoretical description of directive force, establishing dassBNC2014 as a resource for advancing corpus-based speech act research and supporting the development of NLP models capable of recognizing directive speech acts in authentic conversational data.
No takes yet. Share an insight, caveat, or question.
Zhou et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: