<?xml version="1.0" standalone="yes"?>
<Paper uid="P95-1018">
  <Title>Getting the message across in RST-based text</Title>
  <Section position="3" start_page="130" end_page="131" type="metho">
    <SectionTitle>
1 Relational Discourse Analysis
</SectionTitle>
    <Paragraph position="0"> Because the recognition of discourse coherence and structure is complex and dependent on many types of non-linguistic knowledge, determining the way in which cues and other linguistic markers aid that recognition is a difficult problem. The study of cues must begin with descriptive work using intuition and observation to identify the factors affecting cue usage. Previous research (Hobbs, 1985; Grosz and Sidner, 1986; Schiffrin, 1987; Mann and Thompson, 1988; Elhadad and McKeown, 1990) suggests that these factors include structural features of the discourse, intentional and informational relations in that structure, givenness of information in the discourse, and syntactic form of discourse constituents.</Paragraph>
    <Paragraph position="1"> In order to devise an algorithm for cue selection and placement, we must determine how cue usage is affected by combinations of these factors. The corpus study is intended to enable us to gather this information, and is therefore conducted directly in terms of the factors thought responsible for cue selection and placement. Because it is important to detect the contrast between occurrence and nonoccurrence of cues, the corpus study must be be exhaustive, i.e., it must include all of the factors thought to contribute to cue usage and all of the text must be analyzed. From this study, we are deriving a system of hypotheses about cues.</Paragraph>
    <Paragraph position="2"> In this section we describe our approach to the analysis of a single speaker's discourse, which we call Relational Discourse Analysis (RDA). Applying RDA to a tutor's explanation is exhaustive, i.e., every word in the explanation belongs to exactly one element in the analysis. All elements of the analysis, from the largest constituents of an explanation to the minimal units, are determined by their function in the discourse. A tutor may offer an explanation in multiple segments, the topmost constituents of the explanation. Multiple segments arise when a tutor's explanation has several steps, e.g., he may enumerate several reasons why the student's action was inemcient, or he may point out the flaws in the student's step and then describe a better alternative. Each segment originates with an intention of the speaker; segments are identified by looking for sets of clauses that taken together serve a purpose.</Paragraph>
    <Paragraph position="3"> Segments are internally structured and consist of a core, i.e., that element that most directly expresses the segment purpose, and any number of contrlbutors, the remaining constituents in the segment each of which plays a role in serving the purpose expressed by the core. For each contributor in a segment, we analyze its relation to the core from an intentional perspective, i.e., how it is intended to support the core, and from an informational perspective, i.e., how its content relates to that of the core. Each segmei,t constituent, both core and contributors, may itself be a segment with a core:contributor structure, or may be a simpler functional element.</Paragraph>
    <Paragraph position="4"> There are three types of simpler functional elements: (1) units, which are descriptions of domain states and actions, (2) matrix elements, which express a mental attitude, a prescription or an evaluation by embedding another element, and (3) relation clusters, which are otherwise like segments except that they have no core:coatributor structure.</Paragraph>
    <Paragraph position="5"> This approach synthesizes ideas which were previously thought incompatible from two theories of discourse structure, the theory proposed by Grosz and Sidner (1986) and Rhetorical Structure Theory (RST) proposed by Mann and Thompson (1988).</Paragraph>
    <Paragraph position="6"> The idea that the hierarchical segment structure of discourse originates with intentions of the speaker, and thus the defining feature of a segment is that there be a recognizable segment purpose, is due to Grosz and Sidner. The idea that discourse is hierarchically structured by palrwise relations in which one relatum (the nucleus) is more central to the speaker's purpose is due to Mann and Thompson. Work by Moore and Pollack (1992) modified the RST assumption that these palrwise relations are unique, demonstrating that intentional and informational relations occur simultaneously.</Paragraph>
    <Paragraph position="7"> Moser and Moore (1993) point out the correspondence between the relation of dominance among intentions in Grosz and Sidner and the nucleus-satellite distinction in RST. Because our analysis realizes this relation/distinction in a form different from both intention dominance and nuclearity, we have chosen the new terms core and contributor.</Paragraph>
    <Paragraph position="8"> To illustrate the application of RDA, consider the partial tutor explanation in Figure i t. The purpose of this segment is to inform the student that she made the strategy error of testing inside paxt3 too soon. The constituent that expresses the purpose, in this case (B), is the core&amp;quot; of the segment. The other constituents help to achieve the segment purpose.</Paragraph>
    <Paragraph position="9"> We analyze the way in which each contributor relates to the core from two perspectives, intentional and informational, as illustrated below. Each constituent may itself be a segment with its own core:contributor structure. For example, (C) is a subsegment whose tin order to make the example more intelligible to the reader, we replaced references to parts of the circuit with the simple labels partl, part~ and part3.</Paragraph>
    <Paragraph position="10">  purpose is to give a reason for testing part2 first, namely that part2 is more susceptible to damage and therefore a more likely source of the circuit fault. The core of this subsegment is (C.2) because it most directly expresses this purpose. The contributor in (C.1) provides a reason for this susceptibility, i.e.,  that part2 is moved frequently.</Paragraph>
    <Paragraph position="11"> ALTHO A. you know that part1 is good, B. you should eliminate part2 before troubleshooting in part3.</Paragraph>
    <Paragraph position="12"> THIS IS BECAUSE C. 1. part2 is moved frequently</Paragraph>
  </Section>
  <Section position="4" start_page="131" end_page="131" type="metho">
    <SectionTitle>
AND THUS
</SectionTitle>
    <Paragraph position="0"> Due to space limitations, we can provide only a brief description of core:contributor relations, and omit altogether the analysis of the example into the minimal RDA units of state and action units, matrix expressions and clusters. A contributor is analyzed for both its intentional and informational relations to its core. Intentional relations describe how a contributor may affect the heater's adoption of the core. For example, (A) in Figure 1 acknowledges a fact that might have led the student to make the mistake. Such a concession contributes to the hearer's adoption of the core in (B) by acknowledging something that might otherwise interfere with this intended effect. Another kind of intentional relation is evidence, in which the contributors are intended to increase the hearer's belief in the core.</Paragraph>
    <Paragraph position="1"> For example, (C) stands in the evidence relation to (B). The set of intentional relations in RDA is a modification of the presentational relations of RST.</Paragraph>
    <Paragraph position="2"> Each core:contributor pair is also analyzed for its informational relation. These relations describe how the situations referred to by the core and contributor are related in the domain.</Paragraph>
    <Paragraph position="3"> The RDA analysis of the example in Figure 1 is shown schematically in Figure 2. As a convention, the core appears as the mother of all the relations it participates in. Each relation is labeled with both its intentional and informational relation, with the order of relata in the label indicating the linear order in the cliscourse. Each relation node has up to two daughters: the cue, if any, and the contributor, in the order they appear in the discourse.</Paragraph>
  </Section>
  <Section position="5" start_page="131" end_page="132" type="metho">
    <SectionTitle>
2 Reliability of RDA application
</SectionTitle>
    <Paragraph position="0"> To assess inter-coder reliability of RDA analyses, we compared two independent analyses of the same data. Because the results reported in this paper depend only on the structural aspects of the analysis, our reliability assessment is confined to these. The  ure 1 categorization of core:contributor relations will not be assessed here.</Paragraph>
    <Paragraph position="1"> The reliability coder coded one quarter of the currently analyzed corpus, consisting of 132 clauses, 51 segments, and 70 relations. Here we report the percentage of instances for which the reliability coder agreed with the main coder on the various aspects of coding.</Paragraph>
    <Paragraph position="2"> There are several kinds of judgements made in an RDA analysis, and all of them are possible sources of disagreement. First, the two coders could analyze a contributor as supporting different cores. This occurred 7 times (90% agreement). Second, the coders could disagree on the core of a segment. This occurred 2 times (97% agreement). Third, the coders could disagree on which relation a cue was associated with. This occurred 1 time (98% agreement). The final source of disagreement reflects more of a theoretical question than a question of reliable analysis. The coders could disagree on whether a relaturn should be further analyzed into an embedded core:contributor structure. This occurred 8 times (91% agreement).</Paragraph>
    <Paragraph position="3"> These rates of agreement cannot be sensibly compared to those found in studies of (nonembedded) segmentation agreement (Grosz and Hirschberg, 1992; Passonneau and Litman, 1993; Hearst, 1994) because our assessment of RDA reliability differs from this work in several key ways. First, the RDA coding task is more complex than identifying locations of segment boundaries. Second, our subjects/coders are not naive about their task; they are trained. Finally, the data is not spoken as in these other studies.</Paragraph>
    <Paragraph position="4"> Future work will include a more extensive reliability study, one that includes the intentional and informational relations.</Paragraph>
  </Section>
  <Section position="6" start_page="132" end_page="133" type="metho">
    <SectionTitle>
3 Initial results and their application
</SectionTitle>
    <Paragraph position="0"> For each tutor explanation in our corpus, each coder analyzes the text as described above, and then enters this analysis into a database. The technique of representing an analysis in a database and then using database queries to test hypotheses is similar to work using RST analyses to investigate the form of purpose clauses (Vander Linden et al., 1992). Because our analysis is exhaustive, information about both occurrence and nonoccurrence of cues can be retrieved from the database in order to test and modify hypotheses about cue usage. That is, both cue-based and factor-based retrievals are possible. In cue-based retrievals, we use an occurrence of the cue under investigation as the criterion for retrieving the value of its hypothesized descriptive factors. Factor-based retrievals provide information about cues that is unique to this study. In factor-based retrieval, the occurrence of a combination of descriptive factor values is the criteria for retrieving the accompanying cues. In this section, we report two results, one from each perspective: a comparison of the distribution of sn~cE and BECAUSE in our corpus, and the impact of embeddedness on cue selection.</Paragraph>
    <Paragraph position="1"> These results are based on the portion of our corpus that is analyzed and entered into the database, approximately 528 clauses. These clauses comprise 216 segments in which 287 relations were analyzed.</Paragraph>
    <Paragraph position="2"> Accompanying these relations were 165 cue occurrences, resulting from 39 distinct cues.</Paragraph>
    <Section position="1" start_page="132" end_page="133" type="sub_section">
      <SectionTitle>
3.1 Choice of&amp;quot;Since ~' or &amp;quot;Because&amp;quot;
</SectionTitle>
      <Paragraph position="0"> SINCE and BECAUSE were two of the most frequently used cues in our corpus, occurring 23 and 13 times, respectively. To investigate their distribution, we began with the proposal of Elhadad and McKeown (1990). As with our study, their work aims to define each cue in terms of features of the propositions it connects for the purpose of cue selection during text generation. Their work relies on the literature and intuitions to identify these features, and thus provides an important background for a corpus study by suggesting features to include in the corpus analysis and initial hypotheses to investigate.</Paragraph>
      <Paragraph position="1"> Quirk et al. (1972) note several distributional differences between the two cues: (i) since is used when the contributor precedes the core, whereas BECAUSE typically occurs when the core precedes the contributor, (ii) BECAUSE can be used to directly answer a ~#hy question, whereas SINCE cannot, and (iii) BECAUSE can be in the focus position of an it-cleft, whereas SINCE cannot. These distributional differences are reflected in our corpus, and the ordering difference (i) is of particular interest. SINCE and BECAUSE are always placed with a contributor. All but one (22/23) occurrences of Sn~CE accompanied relations in contributor:core order, while all (13/13) occurrences of BECAUSE accompanied relations in core:contributor order 2.</Paragraph>
      <Paragraph position="2"> The crucial factor in distinguishing between S~CE and BECAUSE is the relative order of core and contributor. Elhadad and McKeown (1990) claim that the two cues differ with respect to what Ducrot (1983) calls &amp;quot;polyphony&amp;quot;, i.e., whether the subordinate relatum is attributed to the hearer or to the speaker.</Paragraph>
      <Paragraph position="3"> The idea is that SINCE is used when a relatum has its informational source with the hearer (e.g., by being previously said or otherwise conveyed by the hearer). BECAUSE is monophonous, i.e., its relata originate from a single utterer, while sINCE can be polyphonous. According to Elhadad and McKeown, polyphony is a kind of given-new distinction and thus the ordering difference between the two cues reduces to the well-known tendency for given to precede new. Unfortunately, this characterization of the distinction between s~cg and BECAUSE is not supported by our corpus study.</Paragraph>
      <Paragraph position="4"> As shown in Figure 3, whether or not contributors could be attributed to the hearer did not correlate with the choice of SINCE or BECAUSE. To judge whether a contributor is attributable to the student, mention of ~n action or result of a test that the student previously performed (e.g., you tested 30 to 9round earlier) was counted as 'yes', while information available by observation (e.g., partl a~d part2 are co~r~ected b~l wires), specialized circuit knowledge (e.g., part1 is used bll this test step) and general knowledge (e.g., part~ is more prone to damage ) were counted as 'no'.</Paragraph>
      <Paragraph position="5">  This result shows that the choice between since and BECAUSE is determined by something other than the attributability of contributor to hearer. In future work, we will consider other factors that may determine ordering as possible alternative accounts for this choice. Another factor to be considered in distinguishing the two cues is the embeddedness discussed in the next section. Furthermore, this result demonstrates the need to move beyond small numbers of constructed examples and intuitions formed ~This included answers that begin with BECAUSE. In these cases, we took the core to be the presupposition to the question.</Paragraph>
      <Paragraph position="6">  from unsystematic analyses of naturally occurring data. Only by an exhaustive analysis such as ours can hypotheses such as the one discussed here be systematically evaluated.</Paragraph>
    </Section>
    <Section position="2" start_page="133" end_page="133" type="sub_section">
      <SectionTitle>
3.2 Effect of Segment Embeddedness on
Cue Selection
</SectionTitle>
      <Paragraph position="0"> The second question we report on here concerns whether segment embeddedness affects cue selection.</Paragraph>
      <Paragraph position="1"> Much of the work on cue usage, e.g., (Elhadad and McKeown, 1990; Millis etal., 1993; Schiffrin, 1987; Zukerman, 1990) has focused on pairs of text spans, and this has led to the development of heuristics for cue selection that take into account the relation between the spans and other local features of the two relata (e.g., relative ordering of core and contributor, complexity of each span). However, analysis of our corpus led us to hypothesize that the hierarchical context in which a relation occurs, i.e., what segment(s) the relation is embedded in, is a factor in cue usage.</Paragraph>
      <Paragraph position="2"> For example, recall that the relation between C.1 and C.2 in Figure 2 was expressed as part~ is moved frequently, AND THUS it is more susceptible to damage. Now, the relation between C.1 and C.2 could have been expressed, BECAUSE part2 is muted frequently, it is more musceptible to damage. However, this relation is embedded in the contributor of the relation between B and C, which is cued by THIS IS BECAUSE. Intuitively, we expect that, when a relation is embedded in another relation already marked by BECAUSE, a speaker will select an alternative to BECAUSE to mark the embedded relation. That is, two relations, one embedded in the other, should be signaled by different cues. Because RDA analyses capture the hierarchical structure of texts, we were able to explore the effect of embedding on cue selection. null We hypothesized that cue selection for one relation constrains the cue selection for relations embedded in it to be a different cue. To test this hypothesis, we paired each cue occurrence with all the other cue occurrences in the same turn. Then, for each pair of cues in the same turn, it was categorized in two ways: (1) the embeddedness of the relations associated with the two cues, and (2) whether the two cues are the same, alternatives or different.</Paragraph>
      <Paragraph position="3"> Two cues are alternatives when their use with a relation would contribute (approximately) the same semantic content s . The sets of alternatives in our data are {ALSO,AND}, {BUT,ALTHOUGH,HOWEVER) and SBecause it is based on a test of intersubstitutability, the taxonomy proposed by Knott and Dale (1994) does not establish the sets of alternatives that are of interest here. Two cues may be intersubstitutable in some contexts but not semantic alternatives (e.g., AND and BECAUSE), or they may be semantic alternatives but not intersubstitutable because they are placed in different positions in a relation (e.g., so and BECAUSE).</Paragraph>
      <Paragraph position="4"> {BECAUSE,SINCE,SO,THUS,THEREFOI:tE}. The question is whether the choice between the same and an alternate cue correlates with the embeddedness of the two relations.</Paragraph>
      <Paragraph position="5"> As shown in Figure 4, we can conclude that, when a relation is going to have a cue that is semantically similar to the cue of a relation it is embedded in, an alternative cue must be chosen. Other researchers in text generation recognized the need to avoid repetition of cues within a single text and devised heuristics such as &amp;quot;avoid repeating the same connective as long as there are others available&amp;quot; (Roesner and Stede, 1992). Our results show that this heuristic is over constraining. The first column of Figure 4 shows that the same cue may occur within a single explanation as long as there is no embedding between the two relations being cued. Based on these results, our text generation algorithm will use embeddedness as a factor in cue selection.</Paragraph>
      <Paragraph position="6">  tween same and alternate cues.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>