<?xml version="1.0" standalone="yes"?>
<Paper uid="E99-1029">
  <Title>Parsing with an Extended Domain of Locality</Title>
  <Section position="2" start_page="0" end_page="217" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> One of the most basic properties of Tree Adjoining Grammars (TAGS) is that they have an extended domain of locality (EDOL) (Joshi, 1994). This refers to the fact that the elementary trees that make up the grammar are larger than the corresponding units (the productions) that are used in phrase-structure rule-based frameworks. The claim is that in Lexicalized TAGS (LTAGs) the elementary trees provide a domain of locality large enough to state co-occurrence relationships between a lexical item (the anchor of the elementary tree) and the nodes it imposes constraints on. We will call this the extended domain of locality hypothesis.</Paragraph>
    <Paragraph position="1"> For example, wh-movement can be expressed locally in a tree that will be anchored by a verb of which an argument is extracted. Consequently, features which are shared by the extraction site and the wh-word, such as case, do not need to be percolated, but are directly identified in the tree. Figure 1 shows a tree in which the case feature  are marked with the symbols o, $ and *, respectively. Words in parenthesis are included in trees to provide examples of strings this tree can derive.</Paragraph>
    <Paragraph position="2"> Much of the research on TAGS can be seen as illustrating how its EDOL can be exploited in various ways. However, to date, only indirect evidence has been given regarding the beneficial effects of the EDOL on parsing efficiency. The argument, due to Schabes (1990), is that benefits to parsing arise from lexicalization, and that lexicalization is only possible because of the EDOL. A parser dealing with a lexicalized grammar needs to consider only those elementary structures that can be associated with the lexical items appearing in the input. This can substantially reduce the effective grammar size at parse time. The argument that an EDOL is required for lexicalization is based on the observation that not every set of trees that can be generated by a CFG can be generated by a lexicalized CFG. But does the EDOL have any other more direct effects on parsing efficiency? On the one hand, it is a consequence of the EDOL that wide-coverage LTAGs are larger than their rule-based counterparts. With larger elementary structures, generalizations are lost regarding the internal structure of the elementary trees. Since parse time depends on grammar size, this could have an adverse effect on parsing efficiency. However, the problem of grammar size in TAG has to some extent been addressed both with respect to grammar encoding (Evans et al., 1995; Candito, 1996) and parsing (Joshi and Srinivas, 1994; Evans and Weir, 1998).</Paragraph>
    <Paragraph position="3"> On the other hand, if the EDOL hypothesis holds for those dependencies that are being checked by the parser, then the burden of passing feature values around during parsing will be less than in a rule-based framework. If all dependencies that the parser is checking can be stated directly within the elementary structures of the grammar, they do not need to be computed dynamically during the parsing process by means of feature percolation. For example, there is no need to use a slash feature to establish filler-gap dependencies over unbounded distances across the tree if the EDOL  makes it possible for the gap and its filler to be located within the same elementary structure.</Paragraph>
    <Paragraph position="4"> This paper presents an investigation into the extent to which the EDOL reduces the need for feature passing in two existing wide-coverage grammars: the XTAG grammar (XTAG-Group, 1995), and the LEXSYS grammar (Carroll et al., 1998).</Paragraph>
    <Paragraph position="5"> It can be seen as an evaluation of how well these two grammars make use of the EDOL hypothesis with respect to those dependencies that are being checked by the parser.</Paragraph>
  </Section>
class="xml-element"></Paper>