<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-1032">
  <Title>Correcting ESL Errors Using Phrasal SMT Techniques</Title>
  <Section position="9" start_page="255" end_page="255" type="concl">
    <SectionTitle>
6 Conclusion and Future Directions
</SectionTitle>
    <Paragraph position="0"> In this pilot study we have shown that SMT techniques have potential to provide error correction and stylistic writing assistance to L2 learners.</Paragraph>
    <Paragraph position="1"> The next step will be to obtain a large dataset of pre- and post-editing ESL text with which to train a model that does not rely on engineered data. A major purpose of the present study has been to determine whether our hypothesis is robust enough to warrant the cost and effort of a collection or data creation effort.</Paragraph>
    <Paragraph position="2"> Although we anticipate that it will take a significant lead time to assemble the necessary aligned data, once a sufficiently large corpus is in hand, we expect to begin exploring ways to improve our SMT system by tailoring it more specifically to the demands of editorial assistance. In particular, we expect to be looking into alternative word alignment models and possibly enhancing our system's decoder using some of the richer, more structured language models that are beginning to emerge.</Paragraph>
  </Section>
class="xml-element"></Paper>