<?xml version="1.0" standalone="yes"?>
<Paper uid="P00-1012">
  <Title>References</Title>
  <Section position="8" start_page="0" end_page="0" type="concl">
    <SectionTitle>
5 Conclusion
</SectionTitle>
    <Paragraph position="0"> In this paper, we have presented the results of applying a number of statistical and machine learning techniques to the problem of predicting the order of prenominal adjectives in English. The scores for each of the methods are summarized in table 1. The best methods yield around 90% accuracy, better than the best previously published methods when applied to the broad domain data of the British National Corpus. Note that Mc-Nemar's test (Dietterich, 1998) confirms the significance of all of the differences reflected here (with p&lt;0:005) with the exception of the difference between purely morphological MBL and the method based on positional probabilities.</Paragraph>
    <Paragraph position="1"> From this investigation, we can draw some additional conclusions. First, a solution specific to adjective ordering works better than a general probabilistic filter. Second, machine learning techniques can be applied to a different kind of linguistic problem with some success, even in the absence of syntagmatic context, and can be used to augment a hand-built competence grammar. Third, in some cases statistical and memory based learning techniques can be combined in a way that performs better than either individually.</Paragraph>
  </Section>
class="xml-element"></Paper>