ABSTRACT Intonation and prosody provide information unavailable in orthographic transcriptions, yet crucial to understanding the intended meaning of an utterance and to generating natural- sounding synthetic speech. A full understanding of these phenomena is thus prerequisite to advance in many areas of linguistic research and technology. This understanding requires access to large labeled speech corpora. Over the last year, a diverse group of researchers, from Electrical Engineers and Computer Scientists to Psychologists and Linguists, have cooperated in devising and testing a proposed standard prosodic transcription system for the purpose of annotating large multi- site shared corpora. The next step in this cooperative effort is to hold a workshop (1) to develop training materials for new transcribers and to plan tools for automating aspects of the transcription, (2) to investigate the full range of needs for prosodically annotated speech databases in the speech and natural language research communities while coordinating plans for annotating existing data bases and for designing new ones in view of these needs, and (3) to extend the standard to other dialects of English. The tangible products of the conference will be an online training course and a volume of Conference Notes edited from the reports and discussion at the conference.
