File Information

File: 05-lr/acl_arc_1_sum/cleansed_text/xml_by_section/concl/05/i05-6008_concl.xml

Size: 2,245 bytes

Last Modified: 2025-10-06 13:54:37

<?xml version="1.0" standalone="yes"?>
<Paper uid="I05-6008">
  <Title>Linguistically enriched corpora for establishing variation in support verb constructions</Title>
  <Section position="6" start_page="68" end_page="69" type="concl">
    <SectionTitle>
6 Conclusion and further improvements
</SectionTitle>
    <Paragraph position="0"> The corpus-based method extracts evidence of variation and modification within support verb constructions. The method is sufficiently efficient in extracting proof of morphological productivity, specifier variation and adjectival modification inside LVCs, but at least one instance of each type of variation needs to be manually assessed to determine whether the LVC interpretation is present.</Paragraph>
    <Paragraph position="1"> The evidence retrieved allows us to establish the required syntactic structure, lexical restrictions and furthermore, a preliminary classification of LVCs. Our findings form the basis of the lexical annotation of these expressions in Alpino.</Paragraph>
    <Paragraph position="2"> A few ideas to enhance the method described in order to improve the quality of the retrieved evidence follow. During compilation of the raw subcorpus, we will adapt the method so that, for each P N V triple, all verb and noun variant forms are retrieved from an existing lexicon. This ensures that the 'subcorpus compiler' collects all possible variants from the TwNC. Given that the parsed data includes dependency relations we are trying different methods to infer the complete subcategorization frame of each LVC. So far, an LVC is represented as a P N V triple, but we need to know other syntactic requirements of the predicate. Access to subcategorization frames ought to improve the extraction of variation evidence. Finally, the experiments described concentrate on support verb constructions. It is sometimes difficult to distinguish a support verb construction from an idiomatic expression. Thus, some of the expressions might perfectly belong to the idioms class, rather than the support verb construction group. A related question is how to distinguish  the literal use of triples from the support verb construction use automatically. This still needs a solution.</Paragraph>
  </Section>
class="xml-element"></Paper>
Download Original XML