On identifying basic discourse units in speech: theoretical and empirical issues

In spite of its crucial role in discourse segmentation and discourse interpretation, there is no consensus in the literature on what a discourse unit is and how it should be identified. Working with spoken data, we claim that the basic discourse unit (BDU) is a multi-dimensional unit that should be defined in terms of two linguistic criteria: prosody and syntax. In this paper, we explain which criteria are used to perform the prosodic and syntactic segmentation, and how these levels are mapped onto one another. We discuss three types BDUs (one-to-one, syntax-bound, prosody-bound) and open up a number of theoretical issues with respect to their function in discourse interpretation.