<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-2708">
  <Title>Prague Czech-English Dependency Treebank Any Hopes for a Common Annotation Scheme?</Title>
  <Section position="7" start_page="0" end_page="0" type="concl">
    <SectionTitle>
7 Conclusion
</SectionTitle>
    <Paragraph position="0"> We have described the process of building the first version of a parallel treebank for two relatively distant languages, Czech and English, during which we have also attempted to reconcile two fairly incompatible linguistic theories used for their description.</Paragraph>
    <Paragraph position="1"> The resulting data collection contains data syntactically annotated on several layers of analysis. There have already been experimental machine translation systems MAGENTA (HajiVc et al., 2002) and DBMT ( VCmejrek et al., 2003) confirming the exploitability of the corpus and showing that we are capable of performing automatic transformations from phrase structures to dependency representation with an acceptable, though still not impeccable quality.</Paragraph>
    <Paragraph position="2"> However, for both languages, we have presented examples of phenomena, for which the &amp;quot;native&amp;quot; annotation scheme does not provide a sufficiently fine-grained analysis. In such cases, automatic conversion between annotation schemes is not possible, and the less we can hope for successfull machine translation.</Paragraph>
    <Paragraph position="3"> The question of enhancing the annotation schemes to allow for a lossless transformation between them remains still open, and its difficulty presents a yet unfathomed depth.</Paragraph>
  </Section>
class="xml-element"></Paper>