Michael Wagner



Canada Research Chair in Speech and Language Processing

Tier 2 - 2009-06-01
Renewed: 2014-06-01
McGill University
Social Sciences and Humanities

514-730-0350
chael@mcgill.ca

Coming to Canada from


Cornell University, United States

Research involves


Investigating the production and perception of sentence prosody to understand how language works and how it is processed.

Research relevance


This research may lead to better techniques for speech therapy, language instruction, and speech-recognition systems, as well as more easily understood synthesized speech.

It’s not (just) what you say, it’s how you say it


How we pronounce a sentence can completely change its meaning. For example, you can use the words “I should apologize” to convey your intention to apologize. But if you emphasize “I” with a rising intonation that expresses incredulity, those same words could instead convey your reluctance to apologize, and belief that someone else should.

Our choice of emphasis and intonation—the so-called “prosody” of what we say—is an important but understudied part of what we encode in our speech. For example, the prosody of a sentence often reflects what is new information in a conversation versus what is established background. Prosody also reveals aspects of the sentence’s structure, namely how the words group syntactically and relate semantically. Using an appropriate prosody makes it substantially easier to understand a sentence; using the wrong one can create confusion.

As Canada Research Chair in Speech and Language Processing, Dr. Michael Wagner is trying to determine how prosody works by integrating insights from different research fields. In particular, he uses production and perception experiments in an attempt to answer two key questions: What information is reflected in the prosody of a phrase or sentence, and how much of this information is retrieved by the listener and used in speech processing?

By using sentence prosody as a window into how we process language, Wagner’s research will deepen our understanding of human language. It also has potential implications in language instruction, speech therapy, synthesized speech and speech-recognition systems.