<?xml version="1.0" standalone="yes"?>
<Paper uid="P03-1070">
  <Title>Towards a Model of Face-to-Face Grounding</Title>
  <Section position="3" start_page="0" end_page="0" type="relat">
    <SectionTitle>
2 Related Work
</SectionTitle>
    <Paragraph position="0"> Conversation can be seen as a collaborative activity to accomplish information-sharing and to pursue joint goals and tasks. Under this view, agreeing on what has been said, and what is meant, is crucial to conversation. The part of what has been said that the interlocutors understand to be mutually shared is called the common ground, and the process of establishing parts of the conversation as shared is called grounding [1]. As [2] point out, participants in a conversation attempt to minimize the effort expended in grounding. Thus, interlocutors do not always convey all the information at their disposal; sometimes it takes less effort to produce an incomplete utterance that can be repaired if needs be.</Paragraph>
    <Paragraph position="1"> [3] has proposed a computational approach to grounding where the status of contributions as provisional or shared is part of the dialogue system's representation of the &amp;quot;information state&amp;quot; of the conversation. Conversational actions can trigger updates that register provisional information as shared. These actions achieve grounding. Acknowledgment acts are directly associated with grounding updates while other utterances effect grounding updates indirectly, because they proceed with the task in a way that presupposes that prior utterances are uncontroversial. [4], on the other hand, suggest that actions in conversation give probabilistic evidence of understanding, which is represented on a par with other uncertainties in the dialogue system (e.g., speech recognizer unreliability). The dialogue manager assumes that content is grounded as long as it judges the risk of misunderstanding as acceptable.</Paragraph>
    <Paragraph position="2"> [1, 5] mention that eye gaze is the most basic form of positive evidence that the addressee is attending to the speaker, and that head nods have a similar function to verbal acknowledgements. They suggest that nonverbal behaviors mainly contribute to lower levels of grounding, to signify that interlocutors have access to each other's communicative actions, and are attending. With a similar goal of broadening the notion of communicative action beyond the spoken word, [6] examine other kinds of multimodal grounding behaviors, such as posting information on a whiteboard. Although these and other researchers have suggested that nonverbal behaviors undoubtedly play a role in grounding, previous literature does not characterize their precise role with respect to dialogue state.</Paragraph>
    <Paragraph position="3"> On the other hand, a number of studies on these particular nonverbal behaviors do exist. An early study, [7], reported that conversation involves eye gaze about 60% of the time. Speakers look up at grammatical pauses for feedback on how utterances are being received, and also look at the task. Listeners look at speakers to follow their direction of gaze. In fact, [8] claimed speakers will pause and restart until they obtain the listener's gaze. [9] found that during conversational difficulties, mutual gaze was held longer at turn boundaries.</Paragraph>
    <Paragraph position="4"> Previous work on embodied conversational agents (ECAs) has demonstrated that it is possible to implement face-to-face conversational protocols in human-computer interaction, and that correct relationships among verbal and nonverbal signals enhances the naturalness and effectiveness of embodied dialogue systems [10], [11]. [12] reported that users felt the agent to be more helpful, lifelike, and smooth in its interaction style when it demonstrated nonverbal conversational behaviors.</Paragraph>
  </Section>
class="xml-element"></Paper>