<?xml version="1.0" standalone="yes"?>
<Paper uid="N04-1026">
  <Title>Predicting Emotion in Spoken Dialogue from Multiple Knowledge Sources</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
2 The Dialogue System and Corpus
</SectionTitle>
    <Paragraph position="0"> We are currently building a spoken dialogue tutorial system called ITSPOKE (Intelligent Tutoring SPOKEn dialogue system) (Litman and Silliman, 2004), with the goal of automatically predicting and adapting to student emotions. ITSPOKE uses as its back-end the text-based Why2-Atlas dialogue tutoring system (VanLehn et al., 2002). In ITSPOKE, a student types an essay answering a qualitative physics problem. ITSPOKE then engages the student in spoken dialogue to correct misconceptions and elicit more complete explanations, after which the student revises the essay, thereby ending the tutoring or causing another round of tutoring/essay revision. Student speech is digitized from microphone input and sent to the Sphinx2 recognizer. The most probable transcription output by Sphinx2 is then sent to the Why2-Atlas natural language back-end for syntactic, semantic and dialogue analysis. Finally, the text response produced by the back-end is sent to the Cepstral text-to-speech system, then played to the student through a headphone. ITSPOKE has been pilot-tested and a formal evaluation with students is in progress.</Paragraph>
    <Paragraph position="1"> Our human-human corpus contains spoken dialogues collected via a web interface supplemented with an audio link, where a human tutor performs the same task as ITSPOKE. Our subjects are university students who have taken no college physics and are native speakers of American English. Our experimental procedure, taking roughly 7 hours/student over 1-2 sessions, is as follows: students 1) take a pretest measuring their physics knowledge, 2) read a small document of background material, 3) use the web and voice interface to work through up to 10 problems with the human tutor (via essay revision as described above), and 4) take a post-test similar to the pretest.1 Our corpus contains 149 dialogues from 17 students. An average dialogue contains 45.3 student turns (242.2 words) and 44.1 tutor turns (1096.2 words).</Paragraph>
    <Paragraph position="2"> A corpus example is shown in Figure 1, containing the problem, the student's original essay, and an annotated (Section 3) excerpt from the subsequent spoken dialogue (some punctuation is added for clarity).</Paragraph>
    <Paragraph position="3"> PROBLEM (TYPED): If a car is able to accelerate at 2 m/sa0 , what acceleration can it attain if it is towing another car of equal mass? ESSAY (TYPED): The maximum acceleration a car can reach when towing a car behind it of equal mass will be halved. Therefore, the maximum acceleration will be</Paragraph>
    <Paragraph position="5"> the car such that it accelerates forward?</Paragraph>
    <Paragraph position="7"/>
  </Section>
class="xml-element"></Paper>