<?xml version="1.0" standalone="yes"?>
<Paper uid="I05-2042">
  <Title>Toward a Unified Evaluation Method for Multiple Reading Support Systems: A Reading Speed-based Procedure</Title>
  <Section position="3" start_page="244" end_page="244" type="intro">
    <SectionTitle>
2 The Comprehension-based Methods
2.1 Ohguro (1993)
</SectionTitle>
    <Paragraph position="0"> Ohguro (1993) carried out an experiment in which the efficacy of an English-Japanese w/p-MT-system was examined and reported that a w/p-MT-system would be of more aid to those with a lower reading ability. Fifty-four non-native English speakers took part in the experiment. Ohguro (1993) prepared 28 texts with 80 comprehension questions extracted from various Test of English for International Communication (TOEIC) texts.</Paragraph>
    <Paragraph position="1"> The experiment held two phases. First, all the participants read 14 English-only texts and answered 40 comprehension questions. On the basis of the test score, the participants were divided into two groups so as to balance the reading ability between them. Then, Ohguro (1993) gave English-only texts to one group, the control group, and provided texts supported with the a w/p-MT-system to the other group.</Paragraph>
    <Paragraph position="2"> Ohguro (1993) hypothesized that the control group would get similar test scores on both tests, as opposed to varying test scores from the other group. In addition, it was predicted that the scores of the non-control group would depend on the reading ability of the group members with respect to TOEIC scores (Hypothesis I). That is, a higher test score would be expected for those with a lower TOEIC score group. Thus, Hypothesis I was incorrect given the results. Ohguro (1993) reanalysed the increase in the test scores by dividing that group into two. Under this revised analysis, he hypothesized that a greater increase in score would be shown in the second test by those with lower initial scores (the revised Hypothesis I). This revised hypothesis was correct given the result. Ohguro (1993) concluded that the supporting effect of a w/p-MTsystem was greater for those who had a lower reading ability than those highly skilled readers.</Paragraph>
    <Section position="1" start_page="244" end_page="244" type="sub_section">
      <SectionTitle>
2.2 Fuji et al. (2001)
</SectionTitle>
      <Paragraph position="0"> Fuji et al. (2001) examined how the efficacy of an English-Japanese MT system varied depending on English reading ability. Approximately 200 non-native English speakers participated in the experiment. The participants were divided into 12 groups based on their TOEIC scores.</Paragraph>
      <Paragraph position="1"> The score range was between (i) less than 395 and (ii) more than 900. Fuji et al. (2001) prepared three types of texts. One was an English-only text as a control text, another contained only translated sentences by an MT-system, and the other involved both English texts and the MT-system outputs. Each participant read 14 texts, and answered 40 comprehension questions. null Through this experiment, Fuji et al. (2001) observed that translation-only texts would degrade the test scores for the higher TOEIC score group, while the lower score group exhibited no degrading effect. In addition, they found that English texts with MT-outputs might increase the test scores for the lower score group more greatly than the higher score group.</Paragraph>
      <Paragraph position="2"> With respect to the test completion time, Fuji et al. (2001) observed that an MT-system highly shortened the time for the lower score group relative to the higher score group.</Paragraph>
    </Section>
    <Section position="2" start_page="244" end_page="244" type="sub_section">
      <SectionTitle>
2.3 Summary
</SectionTitle>
      <Paragraph position="0"> Through the surveys of these studies, we were able to confirm that both a w/p-MT-system and an MT-system exhibited greater supporting effects on the lower TOEIC score group than the higher TOEIC score group.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>