<?xml version="1.0" standalone="yes"?>
<Paper uid="C73-1028">
  <Title>Normalization of Natural Language for Information Retrieval, Final Technical</Title>
  <Section position="13" start_page="0" end_page="0" type="concl">
    <SectionTitle>
334 ANNETTE STACHOWITZ
</SectionTitle>
    <Paragraph position="0"> Consider, for example, the German verb einhalten. If its object is cloth or something made of cloth, its translations into English is gather; if its object is a movement, a gesture, etc., its translation is check; with an object like law, contract, date, diet, anything which signifies a rule of some kind, its translation is observe; with path, way, course, etc.</Paragraph>
    <Paragraph position="1"> it is translated as follow. The information which lets us choose the correct one from all these readings must be incorporated in the selection restrictions of the verb and in the feature system of the nouns.</Paragraph>
    <Paragraph position="2"> Currently, our word lists include derivational forms and compounds.</Paragraph>
    <Paragraph position="3"> We are planning a study of these items to determine to which extent their syntactic and semo-syntactic features can be derived from those of the underlying roots and affixes. The development of productive derivational and compounding rule systems would, of course, result in a considerable reduction in the size of the dictionary.</Paragraph>
    <Paragraph position="4"> The lists resulting from our work will be appended to our future progress reports. Two intermediate lists have already appeared in our report on Research in German-English Machine Translation on Syntactic Level of August, 1970 and in Normalization of Natural Language for Information Retrieval, 1971. A comprehensive description of our lexicographic work is contained in our two most recent reports, Development of German-English Machine Translation System, of July 1973 and September 1973.</Paragraph>
    <Paragraph position="5"> We hope that our lexicographic work will not only result in the construction of an MT dictionary but will also provide information which may prove interesting and useful to linguists in general.</Paragraph>
  </Section>
class="xml-element"></Paper>