<?xml version="1.0" standalone="yes"?>
<Paper uid="W02-0103">
  <Title>A Web-based Instructional Platform for Constraint-Based Grammar Formalisms and Parsing</Title>
  <Section position="2" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Feature structures have been used prolifically at every level of linguistic theory, and they form the mathematical foundation of our most comprehensive and rigorous schools of syntactic theory, including Lexical-Functional Grammar and Head-driven Phrase Structure Grammar. This data structure is popular because it shares many properties with the first-order terms of classical logic, and in addition provides named access to substructures through paths of features. Often it also includes a type system reminiscent of the taxonomical classification systems that are widely used in knowledge representation, psychology and the natural sciences.</Paragraph>
    <Paragraph position="1"> For teaching a subject like computational linguistics, which draws on a broad curriculum from many traditional disciplines to audiences with mixed backgrounds themselves, feature-structure-based theoretical and computational linguistics have three important properties. First, they are a mature discipline, in which a great deal of accomplishments have been made over the last 20 years, spanning from empirical and conceptual advances in linguistic theory to its mathematical and computational foundations, to grammar development and efficient processing.</Paragraph>
    <Paragraph position="2"> Second, they are pervasive as an already existing representation standard for many levels of linguistic study. Third, they are transparent, reducing complex theories of grammar to a basic collection of mathematical concepts and algorithms for answering formal questions about those theories. One can address the distinction between descriptions of objects and the objects themselves, the difference between consistency and truth, and what it means for a syntactic theory to be not only elegant but correct in a precise and provable sense.</Paragraph>
    <Paragraph position="3"> The purpose of this paper is to discuss how these three properties can be cast into an instructional setting to arrive at a framework for teaching computational linguistics that highlights the integrated nature and precision with which work in this very heterogeneous discipline can be presented. In principle, the framework we are proposing is open-ended, in the sense that additional modules should be added by students and other researchers, subject to the design principles given in Section 3. We are currently designing three of the core modules for this framework: formal foundations, constraint-based grammar implementation, and parsing.</Paragraph>
  </Section>
class="xml-element"></Paper>