ColloquiaTalk by Gina Levow - Friday, May 4th at 2:30 p.m.

[Margery Ishmael]

Friday, May 4th at 2:30 p.m. in Ryerson 251

Gina-Anne Levow
University of Maryland, College Park

Host: Partha Niyogi

Title: Understanding Spoken Corrections in Human-Computer Dialogue

Abstract: Miscommunication in spoken human-computer interaction is unavoidable.
Ironically, the user's attempts to repair these miscommunications are even
more likely to result in recognition failures, leading to frustrating error
``spirals.''   In this talk I will investigate users' adaptations
  to  recognition errors  made by a spoken language system.  Providing
such a detailed characterization will help to enable speech systems
to identify when a correction is taking place and to more accurately
recognizer the content of correction utterances.

In analyzing over 300 pairs of original and repeat correction utterances,
  matched on speaker and lexical content, we found overall increases in
  utterance and pause duration from original to correction.  In addition,
there were significant decreases in pitch minimum and increases in
the number of final low boundary tones.  These contrasts allow the
training  decision tree classifiers  which distinguish corrections from
  original inputs at 65-77% accuracy depending on the type of correction
and information available.  We further focus on those adaptations --
phonological and durational -- that are most likely to adversely impact
  the accuracy of speech recognizers.  We identify several phonological
shifts from conversational to clear speech style.  We determine that the
  observed durations of spoken user corrections from a field trial represent
  increases over, and divergences from, those predicted by a speech
recognizer's underlying model. Furthermore, words in final position
diverge  significantly more than those in non-final position, due to the
  additional effects of phrase-final lengthening.

Treating systematic prosodic variability in spoken input as a source
of new information allows us to better understand human-computer dialogue.
http://www.umiacs.umd.edu/~gina/cv
*The talk will be followed by refreshments in Ryerson 255*
If you would like to meet the speaker, please send e-mail to 
marge at cs.uchicago.edu


-- 
Margery Ishmael
Department of Computer Science
The University of Chicago
1100 E. 58th Street
Chicago, IL. 60637

Tel. 773-834-8977  Fax. 773-702-8487