\documentstyle[11pt]{article}
%\documentstyle{article}
\pagestyle{empty}

\title{Computational Models for Syntactic Analysis  - Their Fitness for Writing a Computational Grammar for Hebrew}
\author{Shuly Wintner \\ shuly@techunix.bitnet \and Uzzi Ornan \\ ornan@cs.technion.ac.il}
\date{}
\renewcommand{\baselinestretch}{1.2}
\pagestyle{empty}
\begin{document}
\pagestyle{empty}
%\maketitle
\begin{center}
{\Large\bf
Computational Models for Syntactic Analysis - \\
Their Fitness for Writing a Computational \\
Grammar for Hebrew
}
\end{center}
\begin{abstract}
\pagestyle{empty}
In the past ten years many grammatical formalisms have been developed. A
grammatical formalism is a formal tool by which grammars for natural languages
can be written. Most of the formalisms are based on linguistic theories; some
of them were used, together with an implementation of a parsing algorithm, to
produce a {\em parser-generator} -- a computational model for developing
parsers.

In this work we describe and evaluate some of these models, namely PATR,
Generalized LR Parser/Compiler and Slot Grammar. As our main
interest is to devise a computational grammar for Hebrew, our point of view
will be that of the grammar-writer. Our criteria will be the tools with which
the formalisms provide the grammar writer and the ease of writing a grammar
using these tools. 
Among the topics we discuss are the importance and limitations of unification
as an operation on feature structures, the limitations of phrase structure
rules as a framework for describing the grammar of languages with free word
order, the need for special tools to deal with words that do not constitute a lexical
item and other issues.
We'll use examples from a grammar for Hebrew that is currently
being developed. As a consequence we'll try to sketch some of the properties of
a `good' computational model for syntactic analysis.
\end{abstract}

\pagestyle{empty}
\section*{}
\mbox{}
\end{document}