<?xml version="1.0" standalone="yes"?>
<Paper uid="J97-2005">
  <Title>Squibs and Discussions A Delayed Syntactic-Encoding-based LFG Parsing Strategy for an Indian Language Bangla Probal Sengupta* Indian Statistical Institute</Title>
  <Section position="1" start_page="0" end_page="0" type="abstr">
    <SectionTitle>
1. Introduction
</SectionTitle>
    <Paragraph position="0"> In this squib, we propose a technique aimed at efficient computer implementation of LFG-based parsers for Indian languages in general and Bangla (Bengali) in particular.</Paragraph>
    <Paragraph position="1"> (For the LFG formalism, see Kaplan and Bresnan \[1982\].) The technique may also be useful for other languages having similar properties.</Paragraph>
    <Paragraph position="2"> Indian languages are mostly nonconfigurational and highly inflectional. Grammatical functions (GF's) are predicted by case inflections (markers) on the head nouns of noun phrases (NPs) and postpositional particles in postpositional phrases (PPs).</Paragraph>
    <Paragraph position="3"> However, in many cases the mapping from case marker to GF is not one-to-one.</Paragraph>
    <Paragraph position="4"> The classical technique for non-configurational syntactic encoding of GF's (Bresnan 1982b) therefore requires a number of alternations to be thrown in to handle this phenomenon. The resulting nondeterminism in the parser implementation leads to a non-efficient unification component. The problem here, however, is not of unbounded functional uncertainty (described, with proposed solutions, in Kaplan, Maxwell, and Zaenen \[1987\], Kaplan and Maxwell \[1988\], and Kaplan and Zaenan \[1990\]), but rather, one of disjunctive constraint satisfaction bounded within the matrix. Disjunctive constraint satisfaction leads to a degradation of efficiency of the unification component of LFG, as has been pointed out in Knight (1989) and Maxwell and Kaplan (1991). 1 A closer look at the languages reveals that most disjunctions do not exist if an a priori knowledge of the verb (which is generally at the end of the sentence, since Indian languages are mostly verb final and the verb is the last lexeme encountered in a left-to-right scan of the parser) is available. Here we propose a technique that uses this fact to reduce alternations in syntactic encoding. Our method is based on a delayed evaluation of syntactic encoding schema. We treat the points of syntactic encoding of noun phrases as forward references that are temporarily maintained in a symbol table for later binding. A new metavariable, augmentation of the scope of the Locate operator, and a special type of schema (called m-structure) to be projected by the verb are some of the salient features of our technique.</Paragraph>
  </Section>
class="xml-element"></Paper>