<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-0403">
  <Title>Numbat: Abolishing Privileges when Licensing New Constituents in Constraint-oriented Parsing</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> In linguistics, the term gradience is often used to refer to the notion of acceptability as a gradient, as opposed to a more classical all-or-none notion.</Paragraph>
    <Paragraph position="1"> The research goal of this project is to build an experimental platform for computing gradience, i.e.</Paragraph>
    <Paragraph position="2"> for quantifying the degree of acceptability of an input utterance. We called this platform Numbat.</Paragraph>
    <Paragraph position="3"> In order to be able to quantify such a gradient of acceptability with no a priori opinion on the influence played by different types of linguistic relationships, we want to adopt a framework where no one type of (syntactic) relation (e.g. dependency, immediate dominance, or linear precedence) is preferred over the other ones. Although a constraint-oriented (CO) paradigm such as Prop-erty Grammars (Blache, 2001) theoretically does not rely on any preferred relations, we observe that the parsing strategies implemented so far (Morawietz and Blache, 2002; Balfourier et al., 2002; Dahl and Blache, 2004; VanRullen, 2005) do not account for such a feature of the formalism. The strategy we have designed overcomes that problem and allows for constituents to be licensed by any type of relation. Not only does our approach maintain a close connection between implementation and underpinning theory, but it also allows for the decisions made with respect to gradience to be better informed. The purpose of the present paper is to present this new parsing strategy, and to emphasise how it &amp;quot;abolishes the privilege&amp;quot; usually only granted to a subset of syntactic relationships.</Paragraph>
    <Paragraph position="4"> Section 2 presents some background information about the CO approaches and briefly introduces the Property Grammars formalism. Section 3 exposes and discusses the parsing strategy implemented in Numbat. Section 4 then draws the conclusion.</Paragraph>
  </Section>
class="xml-element"></Paper>