<?xml version="1.0" standalone="yes"?>
<Paper uid="P96-1039">
  <Title>An Information Structural Approach to Spoken Language Generation</Title>
  <Section position="7" start_page="299" end_page="299" type="concl">
    <SectionTitle>
5 Conclusions
</SectionTitle>
    <Paragraph position="0"> The generation architecture described above and implemented in Quintus Prolog produces paragraphlength, spoken monologues concerning objects in a simple knowledge base. The architecture relies on a mapping between a two-tiered information structure representation and intonational tunes to produce speech that makes appropriate contrastive distinctions prosodically. The process of natural language generation, in accordance with much of the recent literature in the field, is divided into three processes: high-level content planning, sentence planning, and surface generation. Two points concerning the role of intonation in the generation process are emphasized. First, since intonational phrasing is dependent on the division of utterances into theme and theme, and since this division relates consecutive sentences to one another, matters of information structure (and hence intonational phrasing) must be largely resolved during the high-level planning phase. Second, since accentual decisions are made with respect to the particular linguistic realizations of discourse properties and entities (e.g. the choice of referring expressions), these matters cannot be fully resolved until the sentence planning phase.</Paragraph>
  </Section>
class="xml-element"></Paper>