<?xml version="1.0" standalone="yes"?>
<Paper uid="E06-1012">
  <Title>Statistical Dependency Parsing of Turkish</Title>
  <Section position="6" start_page="93" end_page="95" type="evalu">
    <SectionTitle>
6 Discussions
</SectionTitle>
    <Paragraph position="0"> Our results indicate that all of our models perform better than the 3 baseline parsers, even when no contexts around the dependent and head units are used. We get our best results with Model 3, where IGs are used as units for parsing and contexts are comprised ofwordfinalIGs. Thehighest accuracy in terms of percent of correctly extracted IG-to-IG relations excluding punctuations (73.5%) was obtained when one word is used as context on both sides of the the dependent.11 We also noted that using a smaller treebank to train our models did not result in a significant reduction in our accuracy indicating that the unlexicalized models are quite effective, but this also may hint that a larger treebank with unlexicalized modeling may not be useful for improving link accuracy.</Paragraph>
    <Paragraph position="1"> A detailed look at the results from the best performing model shown in in Table 4,12 indicates that, accuracy decrases with the increasing sentence length. For longer sentences, we should employmoresophisticated modelspossibly including lexicalization.</Paragraph>
    <Paragraph position="2"> A further analysis of the actual errors made by the best performing model indicates almost 40% of the errors are &amp;quot;attachment&amp;quot; problems: the dependent IGs, especially verbal adjuncts and arguments, link to the wrong IG but otherwise with the samemorphological features asthecorrect one except for the root word. This indicates wemay have to model distance in a more sophisticated way and 11We should also note that early experiments using different sets of morphological features that we intuitively thought should be useful, gave rather low accuracy results.</Paragraph>
    <Paragraph position="3"> 12These results are significantly higher than the best base-line (rule based) for all the sentence length categories.  The Context column entries show the context around the dependent and the head unit. Dl=1 and Dr=1 indicate the use of 1 unit left and the right of the dependent respectively. Hl=1 and Hr=1 indicate the use of 1 unit left and the right of the head respectively. Both indicates both head and the dependent have 1 unit of context on both sides.  perhaps use alimited lexicalization such asincluding limited non-morphological information (e.g., verb valency) into the tags.</Paragraph>
  </Section>
class="xml-element"></Paper>