<?xml version="1.0" standalone="yes"?>
<Paper uid="W03-1706">
  <Title>The Effect of Rhythm on Structural Disambiguation in Chinese</Title>
  <Section position="2" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Syntactic research indicates that prosodic features, including stress, rhythm, intonation, and others, have an impact on syntactic structure. For example, normally in a coordination construction like &amp;quot;A and B&amp;quot;, A and B are interchangeable, that is to say, you can say &amp;quot;B and A&amp;quot; and the change of word order does not change the meaning. However, sometimes A and B are not interchangeable. Quirk et al.(1985) gives the following examples: man and woman * woman and man ladies and gentleman *gentleman and ladies Obviously, the examples above cannot be explained by gender preference. A reasonable explanation is that the length of the words (perhaps in syllables) is playing a role; the first constituent tends to be shorter than the second constituent.</Paragraph>
    <Paragraph position="1"> This feature of the length in syllables of a constituent plays an even more important role in Chinese syntax than in English (Feng, 2000). For example, in the verb-object construction in Chinese, there is a preference for the object to be equal to or longer than the verb. Thus while both &amp;quot;Chong &amp;quot;(plant) and &amp;quot;Chong Zhi &amp;quot;(plant) are verbs and have the same meaning, &amp;quot; Chong /plant Shu /tree&amp;quot; is grammatical while &amp;quot; Chong Zhi /plant Shu /tree&amp;quot; is ungrammatical. However, both verbs allow bi-syllabic nouns as objects (e.g., &amp;quot;Guo Shu &amp;quot;(fruit tree), &amp;quot;Mian Hua &amp;quot;(cotton) etc.). The noun phrases formed by &amp;quot;noun + verb&amp;quot; give us another example in which rhythm feature places constraints on syntax, as indicated in the following examples (ungrammatical with *): Mian Hua /cotton Chong Zhi /planting *Mian Hua /cotton Chong /planting *Hua /flower Chong Zhi /planting *Hua /flower Chong /planting &amp;quot;Mian Hua /cotton Chong Zhi /planting&amp;quot; is grammatical but &amp;quot;Mian Hua /cotton Chong /planting&amp;quot; , &amp;quot;Hua /flower Chong Zhi /planting&amp;quot; and &amp;quot;Hua /flower Chong /planting&amp;quot; are all ungrammatical, although &amp;quot;Mian Hua /cotton&amp;quot; and &amp;quot;Hua /flour&amp;quot; , &amp;quot;Chong Zhi /planting&amp;quot; and &amp;quot;Chong /planting&amp;quot; have the same POS and the same or similar meaning.</Paragraph>
    <Paragraph position="2"> The only difference lies in that they have different number of syllables or different length.</Paragraph>
    <Paragraph position="3"> This paper systematically surveys the effect of rhythm on Chinese syntax from the statistical data from a shallow tree bank. Based on the observation that rhythm places constraints on syntax in Chinese, we try to deploy a feature based on rhythm to improve disambiguation in a probabilistic parser by mixing the rhythm feature into a statistical parsing model.</Paragraph>
    <Paragraph position="4"> The rest of the paper is organized as follows: we present specific statistical analyses of rhythm feature in Chinese syntax in Section 2. Section 3 introduces the content chunk parsing which is the task in our experiment. Section 4 presents the statistical model used in our experiment in which a probabilistic rhythm feature is integrated. Section 5 gives the experimental results and finally Section 6 draws some conclusions.</Paragraph>
  </Section>
class="xml-element"></Paper>