6.911 | January IAP 2006 | Undergraduate

Transcribing Prosodic Structure of Spoken Utterances with ToBI

Syllabus

Course Meeting Times

Lectures: 7 sessions for 4 weeks, 2 hours / session

Course Description

This course presents a tutorial on the ToBI system, for labelling certain aspects of prosody in Mainstream American English (MAE-ToBI). Prosody is the aspect of human speech that is suprasegmental. It includes the phrasing and prominence patterns that are used in addition to the sequence of words to convey meaning in speech. This course provides an introduction to the MAE-ToBI system for transcribing phrasing and prominence patterns. It is designed to familiarize the new labeller with the inventory of phonological tone targets via listening to, labelling and producing spoken examples.

MAE-ToBI was devised by linguists, cognitive scientists, computer scientists and engineers, and has become a widely accepted prosodic labelling system for mainstream American English. In addition, the ToBI approach has been extended to other languages, such as Korean, Japanese, German, Spanish and other dialects of English. As its use has expanded and demand for prosodically labelled speech databases has increased, the need for revised training materials has become apparent.

Course Organization

The course is organized as a free-standing tutorial in a series of sections. After introducing an overview of this approach to prosody and prosodic theory, subsequent sections introduce ToBI labels, in order of increasing complexity. Each section contains audio examples throughout and exercises at the end. Students are expected to study and understand one section each week and to complete the exercises before going on to the next section. Completed exercises which include practice with producing selected contours, will be submitted via sftp to a server.

Table of Contents

1.0 Introduction, Overview and Some Basics
2.0 The ToBI Tutorial
2.1 A First Example, Introducing the Labels: H*, L-L%, 0, 1, and 4
2.2 Adding to the Inventory: L* and H-H%
2.3 Two Accents in an Intonational Phrase
2.4 Other Full Intonational Phrase Boundary Tone Combinations: H-L% and L-H%
2.5 Pitch Accents with more than One Tone: Bi-tonal Accents L+H* and L*+H
2.6 Relationships among Pitch Accents within a Phrase: !H*
2.7 More on Downstep: Bi-tonal Pitch Accents with a Downstepped H Element (!H)
2.8 Intermediate Intonational Phrases: 3 Breaks and Phrase Accents
2.9. The Downstepped High Phrase Accent: !H-L% and !H-
2.10 2 Breaks (beginning, but later talk about it w/ problems, below)
2.11 Labelling Disfluency: The p Diacritic (and the %r Tone Label)
2.10 Break Index 2
2.12 Uncertainty in Labelling
Appendix 1: Straight-line Approximations for Final-pitch-accent/Phrase-tone/Boundary-tone Combinations
Appendix 2: Praat Script to Load Multiple Files

Tutorial Authors

Alejna Brugos, Stefanie Shattuck-Hufnagel, Ph.D., Nanette Veilleux, Ph.D.

Acknowledgements

Preparation of this tutorial benefited substantially from the contributions of many people, including the authors of the ToBI Labelling Guide, Mary Beckman and Gail Elam, from which large sections were taken nearly verbatim; the participants in the NSF-sponsored Workshop on ToBI-labelling of Spontaneous Speech, held at Simmons College in August 2004, including Adam Albright, Mary Beckman, Stefan Benus, Jason Brenier, Allison Blodgett, Jennifer Cole, Gail Elam, Julia Hirschberg, Sun-Ah Jun, Heejin Kim, Sahyang Kim, Laurie Maynell, Akui Nti, John Pitrelli, Shari Speer, Lisa Selkirk, Marc Swerts, Jennifer Venditti, Yelena Yasinnik, and Tae-Jin Yoon; consultants who labeled potential examples and exercises, including Mary Beckman, Laurie Maynell, Allison Blodgett and Jennifer Venditti; Julia Hirschberg, who originated the matrix of straight-line approximations for nuclear-pitch-accent/phrase-tone/boundary-tone combinations, of which an expanded version (created with advice from Mary Beckman) is found in the Appendix; Edward Flemming, who contributed the Praat script that formed the basis for the multi-file-loading tool included in the Appendix; the Linguistics Data Consortium, Ohio State University and The Teaching Company for many utterances used in the examples and exercises; our subjects for generously donating their recorded voices and our sponsors: the National Science Foundation ( NSF 0345627), the National Institutes of Health (NIH RO1 DC 02978 and NIH RO1 DC00075), the MIT Undergraduate Research Opportunities Program, MIT’s Department of Electrical Engineering and Computer Science, and Simmons College.

Learning Resource Types
Online Textbook
Lecture Notes
Activity Assignments