<?xml version="1.0" standalone="yes"?>
<Paper uid="C02-2017">
  <Title>Robust Interpretation of User Requests for Text Retrieval in a Multimodal Environment</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
2 Empirical evidence and user
</SectionTitle>
    <Paragraph position="0"> experiments In order to assess user behaviour, we carried out Wizard-of-Oz experiments (Fraser and Gilbert, 1991). Speech recognition and text retrieval were simulated. In different sessions the users interacted with a number of versions of the system: single input mode versions and versions with combinations of input modes. Their performance in terms of number of interactions as well as task completion time was measured, and their comments regarding the interface and the (simulated) system were collected in a questionnaire. Users were grouped according to previous experience with search engines and the web in general. Our results show that both, beginners and advanced users, preferred multimodal interaction over single input modes, and beginners in particular were able to speed up task completion times significantly with the help of a combination of spoken and written input with mouse clicks (Klein et al., 2001).</Paragraph>
    <Paragraph position="1"> From these experiments, we also obtained a corpus of written and spoken utterances which were considered in the further design of the system. The queries which were posed by the users in spoken language were recorded. The recorded utterances were later read to a speech recognition system. This gave us an impression of the number and type of errors to be expected in dealing with queries in spontaneous speech.</Paragraph>
  </Section>
class="xml-element"></Paper>