<?xml version="1.0" standalone="yes"?>
<Paper uid="I05-6007">
  <Title>Syntactic Identification of Attribution in the RST Treebank</Title>
  <Section position="6" start_page="59" end_page="59" type="concl">
    <SectionTitle>
6 Conclusion and Future Work
</SectionTitle>
    <Paragraph position="0"> We have shown that Attribution relations can be identified successfully by using the syntactic structure of the Penn Treebank. In a sense, then, notating Attribution relations in syntactically parsed texts adds no information. Our hypothesis is that all intra-sentential relations in the RST Treebank are of this character.</Paragraph>
    <Paragraph position="1"> This is important for several reasons. First, it is clear that the relations across sentences in the RST Treebank are not directly derivable from syntax, at least not in any obvious way. Our approach to identifying Attributions is a direct implementation of the description in the RST Tree-bank tagging manual. For inter-sentential relations such as CONTRAST or EXPLANATION-EVIDENCE, the situation is quite different. Syntactic criteria are relevant, but clearly not decisive, as can be observed in (Marcu and Echihabi, 2002). Finally, the elimination of intra-sentential relations like Attribution would appear to be more in line with the original vision behind RST; for example, according to Mann &amp; Thompson (1988), the basic unit for RST relations is the clause.</Paragraph>
  </Section>
class="xml-element"></Paper>