Introduction

Robust, accurate parsing of unrestricted natural language text remains an elusive goal of computational linguistics despite several decades of intense research effort. Conventional parsers utilising hand-crafted generative grammars and knowledge bases require considerable linguistic expertise and knowledge engineering effort to produce, and suffer from problems of under-generation, brittle behaviour and domain dependence. This has lead to a growth of interest in techniques for automatically acquiring syntactic knowledge and robust parsing heuristics directly from language data.

The aim of the workshop was to bring together researchers investigating both practical and theoretical issues in the acquisition of syntactic competence (grammars) and performance (parsers) from language data. Suggested topics of interest at the workshop included:

• statistical and symbolic approaches to grammatical inference and automata induction;
• machine learning of grammars, parsing strategies and heuristics;
• theoretical or empirical studies of approaches to the acquisition of syntactic knowledge or processing models.
• empirical methods in grammar development and parsing (e.g. the acquisition of grammars or subcategorization frames from treebanks).

Bill Keller Saarbruecken, August 1998.