<?xml version="1.0" standalone="yes"?>
<Paper uid="N03-3003">
  <Title>Language choice models for microplanning and readability</Title>
  <Section position="8" start_page="0" end_page="0" type="concl">
    <SectionTitle>
4 Conclusions and future work
</SectionTitle>
    <Paragraph position="0"> This paper described how we used the results of a corpus analysis to build language choice models for a microplanner. We discussed the creation of constraint satisfaction problem graphs for our default &amp;quot;good reader&amp;quot; models and how we adapted the models for poor readers. Our &amp;quot;poor reader&amp;quot; models are based on psycholinguistic evidence, including evidence from our own preliminary experiments. We discussed some of the compromises involved in generating more readable textual output and the impacts that further development could have on GIRL's architecture.</Paragraph>
    <Paragraph position="1"> Plans for future work include expanding the size of our corpus analysis and automating at least some of the analysis and data reconfiguration. We plan further development of the microplanner to prevent incoherent solutions being generated. Further on, we plan to take discourse coherence considerations into account.</Paragraph>
    <Paragraph position="2"> We have plans to carry out additional reading experiments with good and bad readers to investigate whether the constraints we tighten to adjust the language models for poor readers actually produce more readable results. We will generate texts under the default &amp;quot;good reader&amp;quot; models and under the constrained, poor reader, models. We will measure reading speeds and comprehension as in our preliminary experiment.</Paragraph>
    <Paragraph position="3"> (Williams et al. 2003). We predict that, as we found then, good readers will perform equally well on both models and poor readers will perform better on the constrained models. We will also carry out user satisfaction evaluations and carry out evaluation surveys with professional basic skills (adult literacy) tutors.</Paragraph>
  </Section>
class="xml-element"></Paper>