Prosody and Prosodic Models
ICSLP 2002 - September 16, 2002, Denver Colorado
Chilin Shih and Greg Kochanski

Overview

The tutorial will summarize the state of the art of prosody research with an emphasis on intonation (f0). We will present it in a data- and knowledge- driven fashion, starting with speech data and physiology, and building towards quantitative models. We will cover a wide range of intonation phenomena, discussing the variations, the cause of variations, and their implications to prosodic modeling.

Section 1: What is Prosody

The physical bases of prosody: what gestures and acoustic features carry information, and how acoustic measurements relate to physiology.

Prosody as a communications channel: We will discuss what information is carried by prosody and show examples (from several languages) with variations. Linguistic theories will be discussed where applicable.

Prosody used to convey lexical meaning: Accents and tones.
Prosody used to convey non-lexical information: Intonation type (Question vs. declarative sentences).
Prosody used to convey discourse functions: Focus, prominence, discourse segments, etc.
Prosody used to convey emotion.
Prosody tied to the physical system: declination.

Section 2: Prosodic Models

We will comment on the strengths and limitations of a variety of theories and modeling techniques.

If prosody carries meaning, emotion, and information,
- How is all this information carried on a shared channel?
- How is it encoded?
- How much prosody can be predicted from text?
A survey of intonation models.

Section 3: A Modeling Example

We show an example of modeling intonation. In the example, we build a complete, quantitative model of a set of "Confirming Questions" in English. These are questions which implicitly ask whether a digit in a number was heard correctly. The experiment studies the interaction of question, stress, and emphasis.

Bibliography

[ Papers | Top | Stem-ML modeling ]

Greg Kochanski: [ Home ]
Chilin Shih: [ Home ]