<?xml version="1.0" standalone="yes"?>
<Paper uid="P98-2210">
  <Title>Idiomatic object usage and support verbs</Title>
  <Section position="1" start_page="0" end_page="0" type="abstr">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Every language contains complex expressions that are language-specific. The general problem when trying to build automated translation systems or human-readable dictionaries is to detect expressions that can be used idiomatically and then whether the expressions can be used idiomatically in a particular text, or whether a literal translation would be preferred. It follows from the definition of idiomatic expression that when a complex expression is used idiomatically, it contains at least one element which is semantically &amp;quot;out of context&amp;quot;. In this paper, we discuss a method that finds idiomatic collocations in a text corpus. The method detects semantic asymmetry by taking advantage of differences in syntactic distributions.</Paragraph>
    <Paragraph position="1"> We demonstrate the method using a specific linguistic phenomenon, verb-object collocations. The asymmetry between a verb and its object is the focus in our work, and it makes the approach different from the methods that use e.g. mutual information, which is a symmetric measure.</Paragraph>
    <Paragraph position="2"> Our novel approach differs from mutual information and the so-called t-value measures that have been widely used for similar tasks, e.g., Church et al. (1994) and Breidt (1993) for German. The tasks where mutual information can be applied are very different in nature as we see in the short comparison at the end of this paper. The work reported in Grefenstette and Teufel (1995) for finding empty support verbs used in nominallsations is also related to the present work.</Paragraph>
  </Section>
class="xml-element"></Paper>