<?xml version="1.0" standalone="yes"?>
<Paper uid="H92-1086">
  <Title>PROSODIC STRUCTURE, PERFORMANCE STRUCTURE AND PHRASE STRUCTURE</Title>
  <Section position="7" start_page="426" end_page="427" type="concl">
    <SectionTitle>
5. DATA
</SectionTitle>
    <Paragraph position="0"> To compare the models, I examined two sets of data: performance structure data reported by Grosjean, Grosjean and Lane \[7\]; and a set of sentences with hand-marked prosodic boundaries, kindly provided by Julia Hirschberg of AT&amp;T Bell Laboratories.</Paragraph>
    <Paragraph position="1"> Grosjean, Grosjean and Lane conducted two experiments, one examining pauses when subjects read sentences at various speeds, and one examining parsing by linguistically-naive subjects. They report only the data on the pausing experiment, though they claim that the parsing data is highly correlated with the pausing data.</Paragraph>
    <Paragraph position="2"> The data consists of 14 sentences, containing 55 oppor- null tunities for inversions. (An opportunity for inversion is a boundary that, according to the model, is locally more prominent than at least one other boundary). In 52 cases the model makes the correct prediction (5% error).</Paragraph>
    <Paragraph position="3"> The three inversions all involved unexpectedly prominent boundaries around multisyllabic pre-head modifiers at sentence end, hence they arguably reflect a single un* modelled effect. Using the standard measure gives us 42 inversions out of 102 opportunities for inversion, or 41% error, dramatically worse than the licensing measure's 5% error rate. (There are more opportunities for inver: sion because the standard model typically makes more distinctions in boundary prominence.) The second data set consists of 127 sentences from the Darpa ATIS task, with prosodic boundary markings added by Julia Hirschberg. She distinguished three boundary strengths: strong, weak, and no boundary.</Paragraph>
    <Paragraph position="4"> A complication in the prosodic data is the presence of hesitation pauses, which I do not expect a syntactic model to capture. As a primitive expedient, I formulated a rule that I could apply mechanically to distinguish hesitation pauses from &amp;quot;genuine&amp;quot; prosodic boundaries, and I eliminated those boundaries that were hesitation pauses according to the rule. Namely, I eliminated any prosodic boundary immediately following a preposition, conjunction, infinitival to, or a prenominal modifier.</Paragraph>
    <Paragraph position="5"> After eliminating hesitation pauses, I applied the licensing-structure measure and the standard measure.</Paragraph>
    <Paragraph position="6"> Using the licensing measure, there were 363 opportunities for inversions, and 12 observed (3% error). Applying the standard model to 16 sentences drawn at random from the data gives 38 inversions out of 114 opportunities, or 33% error.</Paragraph>
    <Paragraph position="7"> Caution is in order in interpreting these results, in that I have not controlled for all factors that may be relevant. For example, the standard measure generally has a greater range of distinctions in boundary prominence, and that may lead to a larger proportion of errors. Also, the method I use to eliminate hesitation boundaries may help the chunks-and-dependencies model more than it helps the standard model. In short, these are exploratory, rather than definitive results. Nonetheless, they strongly suggest that the chunks-and-dependencies model corresponds to empirical prominences better than the standard model does, hence that syntactic structure may be a better predictor of prosodic and performance structures than previously thought.</Paragraph>
  </Section>
class="xml-element"></Paper>