<?xml version="1.0" standalone="yes"?>
<Paper uid="W01-1625">
  <Title>Melodic cues to turn-taking in English: evidence from perception</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
2 Data Analysis
</SectionTitle>
    <Paragraph position="0"> Following the method used by Caspers (2001), two complete Map Tasks (approximately 30 minutes of speech) were divided into interpausal units (IPUs, cf. Koiso et al. 1998) using a pause threshold of 100ms. Boundaries were then categorised according to three criteria.</Paragraph>
    <Paragraph position="1"> Firstly, each IPU boundary was identified as either occurring within the turn of the same speaker, (category HOLD) or involving a change of speaker (category CHANGE). Secondly, the text before each IPU boundary was judged for syntactic completion. If the utterance was at least potentially complete at that point the boundary was categorised as syntactically complete, otherwise as 'not complete'. Finally, we selected a set of IPUs according to their final contour, identified in terms of the British system of 'nuclear tones' (fall, rise etc.) and in autosegmental-metrical terms of final pitch accent and subsequent boundary tone.</Paragraph>
    <Paragraph position="2"> (Following Gussenhoven et al. (1999) we included the possibility of a boundary tone that was neither low nor high, transcribed as %.) We identified IPUs ending in one of the following contours: a high rise (H* H%), a high level (H* %), a fall-rise (H*L H%), a fall (H*L L%) and a truncated fall (H*L %).</Paragraph>
  </Section>
class="xml-element"></Paper>