<?xml version="1.0" standalone="yes"?>
<Paper uid="E06-1024">
  <Title>Keeping the initiative: an empirically-motivated approach to predicting user-initiated dialogue contributions in HCI</Title>
  <Section position="7" start_page="190" end_page="191" type="concl">
    <SectionTitle>
5 Conclusions
</SectionTitle>
    <Paragraph position="0"> In this paper, the discourse structure initiated by users in HCI situations has been investigated and the results have been three-fold. The structures initiated in HCI are much flatter than in HHC; no general orientation with respect to the aims of a sub-task are presented to the artificial communication partner, and marking is usually reduced. This needs to be accounted for in the mapping of the task-structure onto the discourse model, irrespective of the kind of representation chosen. Secondly, the contents of clarification subdialogues have also been identified as particularly dependent on recipient design. That is, they concern the preconditions for formulating utterances particularly for the respective hearer. Here, the less that is known about the communication partner, the more needs to be elicited in clarification dialogues: however, crucially, we can now state precisely which kinds of elicitations will be found (cf. Table 2). Thirdly, users have been shown to differ in the strategies that they take to solve the uncertainty about the speech situation and we can predict which strategies they in fact will follow in their employment of clarification dialogues on the basisoftheirinitialinteractionwiththesystem(cf.</Paragraph>
    <Paragraph position="1"> Table 3).</Paragraph>
    <Paragraph position="2"> Since the likelihood for users to initiate such clarificatory subdialogues has been found to be predictable, we have a basis for a range of implicit strategies for addressing the users' subsequent linguistic behaviour. Recipient design has therefore been shown to be a powerful mechanism that, with the appropriate methods, can be incorporated in user-adapted dialogue management design.</Paragraph>
    <Paragraph position="3"> Information of the kind that we have uncovered empirically in the work reported in this paper can be used to react appropriately to the different types of users in two ways: either one can adapt the system or one can try to adapt the user (Ogden and Bernick, 1996). Although techniques for both strategies are supported by our results, in general we favour attempting to influence the user's behaviour without restricting it a priori by means  of computer-initiated dialogue structure. Since the reasons for the users' behaviour have been shown tobelocatedontheleveloftheirconceptualisation of the communication partner, explicit instruction may in any case not be useful--explicit guidance of users is not only often impractical but also is not received well by users. The preferred choice is then to influence the users' concepts of their communication partner and thus their linguistic behaviour by shaping (Zoltan-Ford, 1991). In particular, Schegloff's analysis shows in detail the human interlocutors' preference for those location terms that express group membership. Therefore, in natural dialogues the speakers constantly signal to each other who they are, what the other per-son can expect them to know. Effective system design should therefore provide users with precisely those kinds of information that constitute their most frequent clarification questions initially and in the manner that we have discussed.</Paragraph>
  </Section>
class="xml-element"></Paper>