<?xml version="1.0" standalone="yes"?>
<Paper uid="W98-0801">
  <Title>i~::'~:::i~~i~ ~:i~.!!?.::!~!:!~i,i-~. * .:;:..: ~::&amp;quot;: N~RTEL HORTI'I~ RH TELECOM Recognition of spontaneous speech</Title>
  <Section position="3" start_page="15" end_page="15" type="metho">
    <SectionTitle>
N RTEL
NORTHEPSN TELECOM
</SectionTitle>
    <Paragraph position="0"> * Phrases may frequently be ungrammatical.</Paragraph>
    <Paragraph position="1"> - Sentence fragments, sloppy usage.</Paragraph>
    <Paragraph position="2"> * Examples: - Reserve tickets two people ten o'clock show. - Gotta go now.</Paragraph>
    <Paragraph position="3"> * Natural language may also include idiomatic expressions.</Paragraph>
    <Paragraph position="4"> * Examples: - &amp;quot;That's cool!&amp;quot; likely has nothing to do with temperature.</Paragraph>
    <Section position="1" start_page="15" end_page="15" type="sub_section">
      <SectionTitle>
Norte\[ OpenSpeech
</SectionTitle>
      <Paragraph position="0"> People frequently speak in sentence fragments or use innovative constructions for effect. The meaning is clear, but no self-respecting grammar textbook would permit it. Robust parsing solves many of these cases.</Paragraph>
      <Paragraph position="1"> Although the number of rules in the grammar can be increased to include all of the possible variations, a number of new problems will be introduced: * computational complexity of a large number of rules.</Paragraph>
      <Paragraph position="2"> * maintaining and debugging the rules is very difficult. New rules will likely have unforeseen consequences and may conflict with existing rules.</Paragraph>
      <Paragraph position="3"> * determining and writing all of the rules in the first place is time-consuming. Writing rules is similar to writing a program in any other computer language.</Paragraph>
      <Paragraph position="4"> Unless they are carefully designed, large programs are brittle and difficult to make bullet-proof.</Paragraph>
      <Paragraph position="5"> Natural language also includes many other effects, such as idiomatic expressions and puns. Some of these will change with time. Extracting the meaning from expressions containing these can be difficult. In practice, particularly given the state of today's voice synthesizers, people realize that they are speaking to a machine and will adjust their language accordingly.</Paragraph>
      <Paragraph position="6"> The sentence fragments will remain, but some of these effects will be much rarer than they will be in conversation with another human. Thanks to Star Trek, people are used to computers interpreting idiomatic expressions literally.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>