<?xml version="1.0" standalone="yes"?>
<Paper uid="C94-2159">
  <Title>PAUSE AS A PHRASE DEMARCATOR FOR SPEECH AND LANGUAGE PROCESSING</Title>
  <Section position="3" start_page="0" end_page="987" type="intro">
    <SectionTitle>
1 INTRODUCTION
</SectionTitle>
    <Paragraph position="0"> A spontaneous speech understanding system accepts naturally spoken input and understands its meaning.</Paragraph>
    <Paragraph position="1"> hi such a system, speechprocessing and language processiug must be integrated in a sophisticated manner.</Paragraph>
    <Paragraph position="2"> Itowew:r, the integration is not straightforward, as the two are stndied independently art(/ have different processing units. Moreover, spontaneous speech contains unexpected phenomena, such as hesitations, corrections and fragmentary expressions, which thus far have not been treated in linguistic rules.</Paragraph>
    <Paragraph position="3"> The most significant concern in speech processing is raising the recognition accuracy. For that purpose, applying linguistic information, e.g. using stochastic models\[l l, syntactic rules\[2\], sen,antic intbrmation\[3\] and discourse plan@l\], is most promising. In a recent Japanese speech translation system\[5\] b*lnselsu-based syntactic constraints are successfully applied in the speech processing module\[6\] 1, However, rules reprel A bunsetsu rouglfly corresponds to a phrase and is the next largest unit after the word. The nunfl)er of words in a phrase ranges from I to 14, art(\] the mean numl)er is al)ont 317\]. senting the same constraints cannot be used directly in sentence-based language processing, where the primary concern is to understand sentence meaning. In speech recognition, a sequence of words forms a bunselsu and a set of bunseisus then forms a sentence.</Paragraph>
    <Paragraph position="4"> In language processing, on the other hand, where the sentence is the basic processing unit, treating the main verh aud its complements is usually the core of processing. For the sentence kaigi ni moshikomi tai no desu ga, meauing 'I would like to apply for the conference,' the processing discrepancy is sketched in  Although linguistic rules for speech recognition always cope with uncertain l)honeme hypotheses, they still expect well-fornmd speech input, and this is even more true of linguistic rules in language processing.</Paragraph>
    <Paragraph position="5"> In spontaneous speech, however, there are hesitations, corrections and incomplete utterances which are uot treated in the conventional framework.</Paragraph>
    <Paragraph position="6"> In addressing spontaneous speech understanding, two main prohlems must be solved: the absence of common processing components a~s sketched in Figure 1, and our insufficient knowledge of spontaneous speech features. In this paper, we propose the pause as a phrase demarcator and the interpausal phrase as the basic processing unit. A phrase is naturally demarcated with pauses in spoken language and an interpausal phrase often functions as a meaning unit\[8\]\[9\], in spontaneous speech understanding we must both accept naturally spoken input and understand its lneaning. Use of the pause as a phrase demarcator is advantageous for both of these purposes.</Paragraph>
    <Paragraph position="7"> Further, we investigate several frequent spontaneous  speech fleatures using spontaneous speech data\[10\].</Paragraph>
    <Paragraph position="8"> We then apply tile study to speech recognition. We examine the effect of integrating into syntactic rules pausal phenomena and certain features of spoken language, using 118 test sentences.</Paragraph>
  </Section>
class="xml-element"></Paper>