<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-1614">
  <Title>Computational Mechanisms for Pun Generation</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
2 Motivation
</SectionTitle>
    <Paragraph position="0"> It could be argued that computer modelling of humour is worthwhile because it might shed light on human use of humour, and hence could contribute to a cognitive model of humour. Here we shall leave that aside, and consider a case which has been more explicitly argued: that certain practical computer systems will be more effective, or more pleasant to use, if they display humour.</Paragraph>
    <Paragraph position="1"> It has been claimed for some time that humour enhances communication in various ways. In one study, subjects were persuaded more effectively by material including humour [Lyttle, 2001]. In another, human subjects gave more favourable reports of working with computer systems which employed humour (albeit pre-coded, rather than computergenerated) [Morkes et al., 1999].</Paragraph>
    <Paragraph position="2"> Binsted [1995] argues that a user-interface which used humour would be more congenial to interact with. Stock [Stock, 2002; 2003] suggests that computerised humour will have even wider applicability, in advertising, entertainment and education. Nijholt [2002] points out that if virtual agents are to show rich &amp;quot;personalities&amp;quot; in their interactions with humans, some form of humour is essential. McKay [2002] suggests the use of automated humour in a system for second-language learning, and O'Mara and Waller [2003] propose that machine-assisted communication by those with language disabilities (particularly children) could be helped by some software support for humour.</Paragraph>
    <Paragraph position="3"> So far, no computer system for humour-generation has yet been shown to have these benefits. Nevertheless, for the purposes of this paper we shall assume, from the writings cited above, that a case can be made for the desirability of computer-generated humour in practical applications.</Paragraph>
    <Paragraph position="4"> These authors have argued generally for the practical use of humour, not merely puns. However, the type of humour-generation that is likely to be available in the near future is pun-generation. In the following sections, therefore, we shall focus solely on puns, considering the ways in which they might be generated within a broader NLG system.</Paragraph>
  </Section>
class="xml-element"></Paper>