<?xml version="1.0" standalone="yes"?>
<Paper uid="J04-2001">
  <Title>c(c) 2004 Association for Computational Linguistics Inferable Centers, Centering Transitions, and the Notion of Coherence</Title>
  <Section position="4" start_page="120" end_page="121" type="intro">
    <SectionTitle>
)RETAIN ROUGH SHIFT
</SectionTitle>
    <Paragraph position="0"/>
    <Section position="1" start_page="121" end_page="121" type="sub_section">
      <SectionTitle>
Fais Transitions and Coherence
</SectionTitle>
      <Paragraph position="0"> and will assume that these clauses form a flat, linear sequence of discourse units, such that the centering output of the first clause in the sentence is the input to the next, and so on, in the spirit of Kameyama (1998) and Suri and McCoy (1994).</Paragraph>
      <Paragraph position="1">  Because the notion of backward-looking center will be critical to the discussion that follows, we look at this notion in greater detail here. According to Grosz, Joshi, and Weinstein (1986, quoted in Walker, Joshi, and Prince [1998]), a center is realized in an utterance U if it &amp;quot;is an element of the situation described by U or the semantic interpretation of some subpart of U&amp;quot; (page 4). As Walker, Joshi, and Prince (1998) point out, this covers &amp;quot;pronouns, zero pronouns, explicitly realized discourse entities, and...entities inferable from the discourse situation&amp;quot; (page 4). The definition proposed by Grosz, Joshi, and Weinstein allows inferable entities, that is, entities that are not expressed at the surface level of the utterance or immediately recoverable from the subcategorization properties of the verb (as, for example, zero pronouns are) to constitute centers of an utterance. However, the theory does not make explicit the parameters within which to characterize the class of permissible inferable elements or the constraints on doing so. We will return to this difficulty later.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>