<?xml version="1.0" standalone="yes"?>
<Paper uid="P98-2168">
  <Title>A Computational Model of Social Perlocutions</Title>
  <Section position="3" start_page="1020" end_page="1020" type="metho">
    <SectionTitle>
2 Previous Research
</SectionTitle>
    <Paragraph position="0"> Our work builds on results from three disparate areas: natural language generation (NLG), communication studies, and social psychology.</Paragraph>
    <Paragraph position="1"> The NLG community has focused on a small sub-set of the five generally accepted categories of speech acts (Levinson, 1983):  1. Representatives--statements given as true depictions of the world (e.g., asserting, concluding). null 2. Directives---statements attempting to persuede the hearer to do something (e.g., ordering, advising, warning).</Paragraph>
    <Paragraph position="2"> 3. Commissives----statements that commit the speaker to a course of action (e.g., promising, accepting a request, taking a side).</Paragraph>
    <Paragraph position="3"> 4. Expressives---statements expressing a psychological state (e.g., apologizing, congratulating, condoling).</Paragraph>
    <Paragraph position="4"> 5. Declarations---statements effecting an immediate change in the institutional state of affairs  (e.g., christening, firing from employment). In particular, research in NLG has been limited to one type of representative (i.e., informing) and one type of directive (i.e., requesting), and it has further focused on informing's potential to con~/nee the hearer of some fact and requesting's potential to persuade the hearer to do some action (Allen et al., 1994; Appelt, 1985; Bruce, 1975; Cohen and Perfault, 1979; Hovy, 1988; Perrault and Allen, 1979). As a result, it has largely ignored speech acts in other categories, such as promising, advising, and crediting, as well as their potential perlocutionary effects of creating airnnity between speaker and hearer, securing future favors for the speaker, and so on. In contrast, research in communication studies has explored strategies for persuading, creating affinity, comforting, and many other interpersonal goak (Daly and Wiemann, 1994; Marcu, 1997). For example, the strategies for persuading include not only requesting, but also exchange, ingratiation, and sanctions. However, these efforts have not analyzed these strategies in terms of speech act types and perlocutionary effects so that these strategies might be realieed in computational form.</Paragraph>
    <Paragraph position="5"> Finally, research in social psychology has looked at how personality traits affect interpersonal interaction. For example, Kiesler (1983) formulated general rules for describing how one person expressing one trait (e.g., merciful) can lead to another person expressing a symmetric and complementary trait (e.g., appreciative). Such interaction dyads are directly msppable to the speaker/hearer dyad of speech act theory, and the vocabulary of trait terms and predictive rules suggest one way of lending organization to the great variety of perlocutionary effects. Yet, social psychologists have not mapped their general trait terms to the classes of speech acts that might express these traits.</Paragraph>
    <Paragraph position="6"> What's been lacking is an attempt to integrate the lessons learned from these different research efforts to provide an initial model of social perlocutions; that is, a model that describes how specific speech act types have the potential to produce specific effects in a hearer corresponding to a speaker's social goals, and that is specified formally enough to be used as part of text generation systems.</Paragraph>
  </Section>
  <Section position="4" start_page="1020" end_page="1022" type="metho">
    <SectionTitle>
3 Our Model
</SectionTitle>
    <Paragraph position="0"> There are two key questions to address in forming a computational model of social perlocutions: * What are the possible socially-relevant effects of speech acts? * What are the relationships between different effects?</Paragraph>
    <Section position="1" start_page="1020" end_page="1021" type="sub_section">
      <SectionTitle>
3.1 Social Perlocutionary Effects
</SectionTitle>
      <Paragraph position="0"> We have developed a taxonomy of social perlocutionary effects of speech acts. These effects are defined in terms of mental attitudes of the hearer, following the assumption in speech act theory that all perlocutionary effects follow from the hearer's recognition of the speaker's communicative intent. The taxonomy is:</Paragraph>
      <Paragraph position="2"> Beliefs about speaker's precise communicative content and communicative intent.</Paragraph>
      <Paragraph position="3"> Beliefs about the speaker's intent to benefit or harm the hearer.</Paragraph>
      <Paragraph position="4"> Beliefs about the heater's or speaker's responsibilities (ascribed or undertaken).</Paragraph>
      <Paragraph position="5"> Beliefs about (or, impressions of) the speaker's personality traits.</Paragraph>
      <Paragraph position="6"> The heater's emotions.</Paragraph>
      <Paragraph position="7"> The relationship between the hearer and the speaker.</Paragraph>
      <Paragraph position="8"> 7. The hearer's goals.</Paragraph>
      <Paragraph position="9"> We developed this taxonomy by reviewing the communications studies and social psychology litersture, as we\]\] as by analysing a corpus of letters and enudl messages for their speech acts and most prominent social effects. Prior research on speech acts has largely ignored several of these categories, especially the effects on personality impressions, emotions, and the speaker-hearer relationship.</Paragraph>
    </Section>
    <Section position="2" start_page="1021" end_page="1022" type="sub_section">
      <SectionTitle>
3.2 Relationship Between Social Ef-
fects
</SectionTitle>
      <Paragraph position="0"> This taxonomy is important because there appear to be significant restrictions on the relationships between these different classes of effects.</Paragraph>
      <Paragraph position="1"> Figure 2 shows how these different types of effects are related. The arrows represent potential causal links between effects. These links are potential because there are specific conditions associated with specific effects that dictate whether one effect will cause another.</Paragraph>
      <Paragraph position="2"> Essentially, the effects start with the hearer's recognition and acceptance of a message's content and culminates in changes to hearer goals and the relatiouship between the hearer and the speaker. That is, a speech act directly results in beliefs about the content and intent of utterances and these beliefs indirectly result in changes to goals, emotions, and interpersonal relationships. Specficially, these belief can lead to indirect changes in the heater's belief about the speaker's intent to benefit or harm the hearer, as well as changes to the heater's responsibilities that involve the speaker. In turn, changes in belief about whether the speaker intends to benefit or harm the header can lead to changes in the hearer's goals, the heater's emotions, and the heater's impressions of the speaker's personality traits. Finally,  changes to the hearer's emotions can lead to changes in the hearer's relationship with the speaker.</Paragraph>
      <Paragraph position="3"> Our hypothesis is that Figure 2 provides a framework into all speech acts with social effects can be mapped. To test this hypothesis, we analyzed in detail the relationship between the effects of 40 different types of speech acts, and we successfully placed each into this framework (Pautler, 1999). These speech acts were typical of the letters and messages we collected, and they were representative of four of the five main categories of speech acts.1 Figure 3 is an example, showing these effects for apolo~zing, a Although not shown in Figure 3, the causal relationships between these effects have conditions attached to them. In Figure 3, for example, a condition on an apology leading to the hearer believing the speaker feels regret is that the hearer believes the speaker is sincere and there is an act for which 1We did not represent deelar6tions because we chose to focus C/m acts used in casual, interpersonal interactions rather thA~ acts that were institutionally framed.</Paragraph>
      <Paragraph position="4"> =We do not rl.;,, that the model applies to groups other th~n adu/t Westea'ne~. See Bm'nlund (1989) for comparisons en the use of different speech acts by Americana and Japanese.  an apology is appropriate.</Paragraph>
      <Paragraph position="5"> We draw our terminology for describing specific personality traits (e.g., likeable, conscientious) and emotions (e.g., gratitude, liking) from existing taxonomies (Kiesler, 1983; Ortony et al., 1988).</Paragraph>
      <Paragraph position="6"> Figure 3 shows effects with arrows leading to them from other speech acts, such as praising, warnhag, thanking, and so on. These speech acts are there to illustrate that speech acts are related through a web of interlocking effects. That is, the causal relatiouships between speech acts and effects is manyto-many: a single act can have many different effects and any single effect can be brought about by many different acts. For example, expressing a demand can bring about compliance, anger, or both, and similarly, anger can be caused by a variety of other acts, such as issuing a threat. In Figure 3, both praising and apologizing are examples of acts that can increase the heater's liking for the speaker, and both apologiMng and thanking can lead the hearer to believe the speaker is accountable.</Paragraph>
      <Paragraph position="7"> This large web of relationships between the effects of social speech acts leads to the question: How can we efficiently generate the speech acts we need to achieve an appropriate emotional response in the hearer?</Paragraph>
    </Section>
  </Section>
  <Section position="5" start_page="1022" end_page="1024" type="metho">
    <SectionTitle>
4 A Model Of Letter Genera-
</SectionTitle>
    <Paragraph position="0"> tion To illustrate the power of our model of social perlocution, we have applied it to the task of e-mail generation in a system called LetterGen. The system's primary task is to take a high-level communicative goal (e.g., inform a colleague that one can't attend a meeting) and suggest a set of speech acts to achieve that goal. However, once it has made this suggestion, the system then interact with the user to determine which speech acts will appear in the final message and to acquire any additional bat.kground information needed to iustantiate sentence text templates associated with each speech act.</Paragraph>
    <Paragraph position="1"> In addition to the user's explicit input goal, the system works with a set of &amp;quot;standardn user goals. These goals fall into three classes:  1. Cost avoidance avoiding undesired aspects of a current or incipient situation, such as unwanted social perceptions of oneself.</Paragraph>
    <Paragraph position="2"> 2. Status-quo maintenance ~election of an act because one of its effects would reinforce a desired aspect of the current situation (e..g, offeting to help another person because it would reinforce one's self-image as a generous per- son). 3. Trait-based habit--performing of an act as a  timeworn expression of a personality trait.</Paragraph>
    <Paragraph position="3"> These goals can be thought as a stereotypical model of the user (Chin, 1989). These goals are achieved opportunistically during the process of determining speech acts for the explicitly provided user goal.</Paragraph>
    <Section position="1" start_page="1022" end_page="1023" type="sub_section">
      <SectionTitle>
4.1 A Graph-Based Representation
Of Speech Act Relationships
</SectionTitle>
      <Paragraph position="0"> LetterGen essentially represents the perlocutionary effects of speech acts as a large graph. Figure 4 illustrates a portion of this representation that relates the speech acts of declining, thanking, and apologizing. The nodes of the graph represent various effects, and the unlabled edges represent a causal relationships between two effects. There are also constraints on when edges can be traversed (although hey are  5. If an effect is indexed by a mitigates link, follow the link to the mitigating effect in the other chain. Continue with steps 2 and 3.</Paragraph>
      <Paragraph position="1">  As an example, consider the user's communicstive goal to make the hearer believe that the speaker will not attend. Lettergen traverses the graph downwards to locate the speech act Declining. After determining this speech act, LetterGen then traverses the graph upward, moving through its effects, verifying that none of them conflict with known speaker goals. In this case, one of the effects of Declining conflicts with the speaker's goal that the hearer be. lieves the speaker is polite. At this point, LetterGen generates a new goal to mitigate that effect, and recursively uses its algorithm to locate speech acts to achieve that goal. With the failed goal of being perceived as polite, LetterGen's downward traversal locates Thanking and Apologising as appropriate speech acts to mitigate that failure.</Paragraph>
      <Paragraph position="2"> not shown in this figure). Finally, there are mitigates finks between nodes when two effects are contradictory. null A reasonable view of LetterGen's approach is that there is a acr/pt associated with each speech act that captures the causal chain of effects that potentlally follow from it. While these effects could presumably be determined by reasoning from first principles, these scripts can be viewed as standard methogs of achieving communicative goals, and they are essentially equivalent to the communicative strategies proposed by others (Marco, 1997).</Paragraph>
    </Section>
    <Section position="2" start_page="1023" end_page="1023" type="sub_section">
      <SectionTitle>
4.2 Determining Appropriate Speech
Acts
</SectionTitle>
      <Paragraph position="0"> LetterGen's algorithm for producing a response involves 5 steps:  1. Metch the user's goal to one of the nodes (effects) in the graph.</Paragraph>
      <Paragraph position="1"> 2. From the matching effect, traverse graph finks &amp;quot;downward ~ toward the speech act, checking the conditions on each link.</Paragraph>
      <Paragraph position="2"> 3. For every path that reaches an act by satisfying all conditions along the path, add the act to the new message by instantiating the act's text template.</Paragraph>
      <Paragraph position="3"> 4. Detect undesirable side effects of each added  speech act by traversing all links back &amp;quot;up-</Paragraph>
    </Section>
    <Section position="3" start_page="1023" end_page="1024" type="sub_section">
      <SectionTitle>
4.3 An Alternative To Planning
</SectionTitle>
      <Paragraph position="0"> This approach can be viewed as a form of reactive planning. LetterGen can be viewed as having a simple goal (communicate a particular belief to the hearer), forming a plan (finding a set of speech acts that communicate this belief), analyzing the effects of the plan (looking for user goals that are violated by these speech acts), and opportunistically pursuing new goals (to mitigate these violations).</Paragraph>
      <Paragraph position="1"> LetterGen differs significantly from most other efforts in planning speech acts. These efforts explicitly represent speech acts and their effects as plan operators and attempt to synthesize sequences of operators. Unfortunately, as others have pointed out (Cohen and Levesque, 1980; 1990), plan operators are not a good representation when acts have long chains of effects. That's because each chain that resuits from a given act must be conflated to a fiat list of effects, or each effect must be re-envisioned as an act, with one operator for each effect and appropriate preconditions so the operators can form the appropriate chain.</Paragraph>
      <Paragraph position="2"> LetterGen's approach is most similar to the alternative to planning for speech-modeling proposed by Cohen and Levesque (1980, 1990). Their approach uses a set of inference rules and act type definitions and is explicitly designed to capture sequences of this type,</Paragraph>
      <Paragraph position="4"> where A(d) is an act that communicates propositional content d (definitional content for some act type), which induces effect E1 under conditions cl, which induces effect E2 under conditions c2, and so on.</Paragraph>
      <Paragraph position="5"> This rule formalism is directly mappab\]e to the conditionalised causal relations used in our social perlocutions model, with two exceptions. One is that we capture the rules with an annotated graph structure that makes the connectivity among rules explicit (scripts). The other provide a specialized graph-traversal algorithm that takes advantage of key properties of the graph, which allows us to substitute et~cient graph traversal for generallsed planning. null</Paragraph>
    </Section>
  </Section>
  <Section position="6" start_page="1024" end_page="1024" type="metho">
    <SectionTitle>
5 Implementation
</SectionTitle>
    <Paragraph position="0"> The current implementation contains a very detailed model of speech act effects, containing over 400 effects and constraints. It is able to generate a dozen different types of messages, including initiating or terminating a friendship, applying or resigning from a job, congratulating or consoling someone, accepting or declining an invitation, encouraging or discouraging someone from doing an act, thanking someone, and apologizing to someone. Each of these different message types includes an organizational template that places generated acts in an appropriate order for the task.</Paragraph>
    <Paragraph position="1"> An important part of LetterGen is its interaction with the user. Given a selected message type, LetterGen suggests at least three speech acts for the user to choose from. For example, the thanking message type (i.e., make them believe you feel gratitude) can be instantiated crediting (distributing credit), offering (to repay), as well as an overt expression of gratitude (i.e., thanking). For each act chosen by the user, the system queries the user for the background information needed to instantiate an appropriate text template.</Paragraph>
  </Section>
class="xml-element"></Paper>