<?xml version="1.0" standalone="yes"?>
<Paper uid="H94-1115">
  <Title>COMBINING LINGUISTIC AND STATISTICAL TECHNOLOGY FOR IMPROVED SPOKEN LANGUAGE UNDERSTANDING</Title>
  <Section position="1" start_page="0" end_page="0" type="metho">
    <SectionTitle>
COMBINING LINGUISTIC AND STATISTICAL TECHNOLOGY FOR
IMPROVED SPOKEN LANGUAGE UNDERSTANDING
</SectionTitle>
    <Paragraph position="0"/>
  </Section>
  <Section position="2" start_page="0" end_page="0" type="metho">
    <SectionTitle>
PROJECT GOALS
</SectionTitle>
    <Paragraph position="0"> The goal of this project is to develop technology for spoken language understanding which is highly accurate, robust, and fast:, is easily ported to new domains, environments, and chaJmels, and quickly adapts to new speakers. The system combines the DECIPHER speech recognition system with the Gemini natural language understanding system.</Paragraph>
  </Section>
  <Section position="3" start_page="0" end_page="0" type="metho">
    <SectionTitle>
RECENT RESULTS
</SectionTitle>
    <Paragraph position="0"> SRI has developed a spoken language interface to the Official Airline Guide (OAG). Despite a funding gap for more than four months of the year, substantial improvements have been made in the component technologies. On recent ARPA benchmarks. SRI achieved 5.5% word error on the ATIS speech recognition task, 18.2% utterance error on the natural-language understanding task, and 20.7% utterance error on the spoken-language understanding task. Other recent results include: Investigated several speaker-adaptation algorithms for both native and non-native speakers of English.</Paragraph>
    <Paragraph position="1"> The resulting techniques can match speaker-dependent performance (trained on 650 sentences) using 100 adaptation sentences, and outperforms the speaker-dependent system when more than 100 adaptation sentences are used.</Paragraph>
    <Paragraph position="2"> Developed an approach for constructing acoustic models for telephone applications using high-quality recordings, resulting in a substantial savings in effort when porting the ATIS application to a telephone envkonment.</Paragraph>
    <Paragraph position="3"> Developed methods to discriminate &amp;quot;hesitation&amp;quot; from &amp;quot;end-of-utterance&amp;quot; silent pauses based on durational and f0 correlates of preceding syllables. This can have important implications for the design of end-pointing algorithms.</Paragraph>
    <Paragraph position="4"> * Performed a study of filled pauses which showed that they occur almost exclusively in between words in low-probability word sequences.</Paragraph>
    <Paragraph position="5"> * Improved the modeling of out-of-vocabulary words and word-fragments.</Paragraph>
    <Paragraph position="6"> * Developed a class-trigram grammar for ATIS, resulting in a 30% decrease in word error compared to a word-bigram grammar. Approximately half the improvement was due to the trigrams, and half to the classes.</Paragraph>
    <Paragraph position="7"> * Developed methods to incorporate natural-language constraints supplied by the Gemini parser into the DECIPHER recognition search.</Paragraph>
    <Paragraph position="8"> * Increased the speed of the Gemini parser by a factor of four by improved handling of semantic selectional restrictions.</Paragraph>
    <Paragraph position="9"> * Expanded the scope of the SRI ATIS system for ATIS3.</Paragraph>
    <Paragraph position="10"> * Assumed leadership of the effort to define a semantic evaluation methodology for spoken language systems, working out a detailed framework for annotation of the predicate-argument structure of utterances.</Paragraph>
    <Paragraph position="11"> * Collected a total of 2863 ATIS3 training and test utterances (speech, transcriptions, and log files).</Paragraph>
  </Section>
  <Section position="4" start_page="0" end_page="472" type="metho">
    <SectionTitle>
PLANS FOR THE COMING YEAR
</SectionTitle>
    <Paragraph position="0"> In the following year we plan to continue to explore methods for integrating natural-language constraints into speech recognition systems, develop rapid speaker adaptation methods, improve the portability and scalability of the technology, and complete the development of a telephone-based ATIS system.</Paragraph>
  </Section>
class="xml-element"></Paper>