<?xml version="1.0" standalone="yes"?>
<Paper uid="C96-2111">
  <Title>Top-Down Predictive Linking and Complex-Feature-Based Formalisms</Title>
  <Section position="2" start_page="0" end_page="658" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Complex-feature-based formalisms are understood here as equivalent to unification-based formalisms as exemplified by PATR-II, HPSG, and others (cf Shieber 1986, Carpenter 1992).</Paragraph>
    <Paragraph position="1"> Such formalisms typically include a context-free (CF) base, which allows the use of parsing algorithms designed for CF languages despite the fact that complex-feature-based formalisms are essentially more powerful than CF grammars. However, such an adaptation of CF algorithms involves their extension to possibly infinite nonterminal domains, which, as Shieber (1985) and Haas (1989) have shown, is nontrivial. null Various CF algorithms make use of a binary relation between a goal category and the category of a constituent (phrase or word) which either has just been parsed or is to be parsed next. Different terms have been used to designate this relation; Kay (1980) speaks of reachability, while Pereira/Shieber (1987) and others before them use the term linking for the relation. null Whatever term one takes, an important aspect of the relation is that it can be used to reduce the search space of possible syntactic analyses at an earlier point in parsing and thus serves to improve the efficiency of a parser. Shieber (1985, 1992) follows established terminology in speaking of top-down filtering in connection with the prediction step of the Earley algorithm. His central notion of restriction, whereby a restrictor is a finite subset of the paths specified in a feature structure, is related to the technique we introduce here, since both guarantee the finiteness of an otherwise possibly infinite domain of complex categories, but Shieber's restrictors are specified manually.</Paragraph>
    <Paragraph position="2"> We propose a general algorithmic method of compilation that avoids manual specification.</Paragraph>
    <Paragraph position="3"> The focus of this discussion is on the linking relation used to extend left-corner parsers, rather than on the prediction step of the Earley algorithm as with Shieber, although the results carry over.</Paragraph>
    <Paragraph position="4"> Whereas Shieber et al. (1990) have discussed similar techniques in the context of semantic-head-driven generation, we are concerned here with parsing. We view the linking relation not simply as a filter to increase efficiency within the domain of syntactic analysis--this aspect is stressed by Shieber (1985) and other investigators such as Bouma (1991)--but rather as a device for the top-down predictive instantiation of information, as Shieber et al. (1990) have shown for semantic-head-driven generation. In this paper we are concerned especially with morphosyntactic information and illustrate the relevance of predictive linking for morphological analysis and for the analysis of &amp;quot;unknown&amp;quot; or &amp;quot;new&amp;quot; lexical items.</Paragraph>
  </Section>
class="xml-element"></Paper>