<?xml version="1.0" standalone="yes"?>
<Paper uid="E06-2027">
  <Title>Information structure and pauses in a corpus of spoken Danish</Title>
  <Section position="5" start_page="192" end_page="193" type="concl">
    <SectionTitle>
5 Conclusions and further research
</SectionTitle>
    <Paragraph position="0"> In conclusion, the pilot study shows that words making up the topic or the focus of a sentence do not show a general tendency to be preceded by pauses. However, preliminary results indicate that topics tend to be followed by pauses. Furthermore, words belonging to specific syntatic  categories may have a significantly higher probability to be preceded by a pause than a randomly chosen word. In the corpus we have worked with, these words express the property or semantic type whereby the object in focus can be distinguished from other similar objects. In other words, the system by which Danish speakers use pauses seems sensitive to information structure in a subtle way that, at least as far as focus is concerned, creates boundaries that do not necessarily correspond to those between syntactic constituents. null An interesting issue we haven't yet addressed is whether intra-clausal pauses relate to prosodic phrases, which according to Steedman (2001) correspond to information structural constituents. Since the DanPass annotation also foresees a tier for prosodic phrases, this investigation is possible. Furthermore, we want to test whether there are differences in the way in which different users relate pauses to topic establishment and focusing. We know already now that the percentage of pauses per word varies across speakers, and that speakers' individual pause rates do not vary much depending on the task. The corpus provides a very nice means of studying whether they use pauses for different purposes.</Paragraph>
  </Section>
class="xml-element"></Paper>