<?xml version="1.0" standalone="yes"?>
<Paper uid="W02-1507">
  <Title>A Classification of Grammar Development Strategies</Title>
  <Section position="7" start_page="0" end_page="0" type="concl">
    <SectionTitle>
6 Conclusion
</SectionTitle>
    <Paragraph position="0"> We have proposed a classification of grammar development strategies and have examined the advantages and drawbacks of each of the four approaches. We have explained how &amp;quot;grammar replication&amp;quot; may prove an interesting task to compare different development strategies, and have described how grammar replication is currently being used in the Xtag project at the University of Pennsylvania in order to compare hand-crafted grammars, grammars generated with MetaRules, and grammars generated with a MetaGrammar.</Paragraph>
    <Paragraph position="1"> We have reached the conclusion that of the four grammar development strategies proposed, the most promising one consists in automatically acquiring an abstract level of syntactic representation (such as the MetaGrammar). Future work will consist in pursuing this automatic acquisition effort on the Penn Treebank. In parallel, we are investigating how the abstract level we acquire can be used to generate formalisms other than TAGs (e.g. LFG).</Paragraph>
    <Paragraph position="2"> Aknowledgements: We thank the Xtag group, and more particularly W.</Paragraph>
    <Paragraph position="3"> Schuler and R. Prasad for helpful comments on earlier versions of this work. We also thank B. Crabb'e and B. Gaiff'e for their help with the LORIA MetaGrammar compiler.</Paragraph>
  </Section>
class="xml-element"></Paper>