<?xml version="1.0" standalone="yes"?>
<Paper uid="P98-1007">
  <Title>Tense and Connective Constraints on the Expression of Causality</Title>
  <Section position="5" start_page="52" end_page="53" type="concl">
    <SectionTitle>
4 Conclusion
</SectionTitle>
    <Paragraph position="0"> Summary We summarize our observations in the table 1. We consider in this table all the possible configurations one has when the three following parameters vary.</Paragraph>
    <Paragraph position="1">  1. Order of presentation: el before e2 or the other way around (assuming el is the cause of ee).</Paragraph>
    <Paragraph position="2"> 2. Presence of a connective &amp;quot;donc&amp;quot; or &amp;quot;car&amp;quot;. 7 3. Use of PS or IMP.</Paragraph>
    <Paragraph position="3">  C2: non accomplished causality Among the combinations, some are always possible (which does not mean they always convey causality), some are never possible, that is, either uninterpretable or incompatible with causality. Some are sometimes possible, depending on various constraints as shown in this paper. Notice that we mention in this table some configurations we have not considered so far, namely configurations with an IMP-PS sequence. r As we have already said, we are only concerned in this paper with &amp;quot;donc&amp;quot; and mention &amp;quot;car&amp;quot; only for the sake of completeness.</Paragraph>
    <Paragraph position="4">  We mention them here only for the sake of completeness, since they can never be used to express causality.</Paragraph>
    <Paragraph position="5"> The second column of the table gives the discourse relation associated with each configuration. In some cases, it is a cause relation, either in one direction (result-res) or in the other (explanation-exp). The other cases are compatible with a cause relation, without conveying it, which is noted in the table as &amp;quot;suc&amp;quot; (for temporal succession) or &amp;quot;ntr&amp;quot; (neutral-for ambiguous cases between background or temporal succession). null Conclusion This paper shows that the interaction of constraints coming from tenses and connecti.ves is rather delicate to characterize, even in the limited domain of the expression of causality. It also shows, however, that it is possible to draw from the linguistic characterisation of these enough principles to be able to generate discourses conveying causality with good guaranties on the achieved effect, and control over the influence of tenses often neglected in this respect.</Paragraph>
    <Paragraph position="6"> We are presently studying the treatment of other connectives, and the extension to other tenses.</Paragraph>
  </Section>
class="xml-element"></Paper>