Author

Karen Ward

Date

June 2001

Document Type

Dissertation

Degree Name

Ph.D.

Department

Dept. of Computer Science and Engineering

Institution

Oregon Graduate Institute of Science & Technology

Abstract

As our ability to build robust and flexible spoken-language human-computer interfaces increases, we must consider whether and how we should incorporate various human-human discourse mechanisms into our dialogue models. In this dissertation I examine the use and probe the potential usefulness of one of these, acknowledgment. Acknowledgments signal understanding but not necessarily agreement; they serve to assure the conversants that information has been conveyed successfully. They also play a role in managing turn- taking. Before we can incorporate acknowledgments in human-computer interfaces in an effective manner, there are several basic questions that should be answered. In this dissertation I report on a three-part research program in which I examine the use of acknowledgments in human-computer interaction from several perspectives: Recognizing acknowledgments: How can acknowledgments be recognized using low-level prosodic features and contextual cues? In two studies, I analyze corpora of human conversation for prosodic and contextual cues that might be useful for recognizing that an acknowledgment has occurred. Predicting acknowledgments: Can subjects predict where acknowledgments might occur in human-human dialogue? Two studies probed subjects' ability to determine whether acknowledgments might occur after a turn. Eliciting acknowledgments: Are subjects are willing to use acknowledgments in human-computer interaction? I present and discuss a Wizard-of-Oz study in which subjects could control the presentation of information using either acknowledgments or commands. By combining the three approaches, I was able to probe various aspects of the larger issue of understanding how and whether we should incorporate acknowledgments in spoken-language interfaces. Both the corpus studies and the perceptual studies suggest that dialogue-level context will be more important than local cues both for recognizing and for predicting (or generating) acknowledgment behavior in human-computer interfaces. The Wizard-of-Oz study shows that some subjects are willing to use acknowledgment as a turn-taking mechanism even in a fairly limited interface, although other subjects report resistance to the idea; more study is needed to understand the strength and implications of that resistance.

Identifier

doi:10.6083/M4ZK5DM9

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.