The linguistic string parser

Abstract

The linguistic string parser is a system for the syntactic analysis of English scientific text. This system, now in its third version, has been developed over the past 8 years by the Linguistic String Project of New York University. The structure of the system can be traced to an algorithm for natural language parsing described in 1960. This algorithm was designed to overcome certain limitations of the first parsing program for English, which ran on the UNIVAC 1 at the University of Pennsylvania in 1959. The UNIVAC program obtained one preferred grammatical reading for each sentence; the parsing program and the grammar were not separate components in the overall system. The 1960 algorithm obtained all valid parses of a sentence; it was syntax-driven by a grammar consisting of elementary linguistic strings and restrictions on the strings (described below). Successive implementations were made in 1965, in 1967, and in 1971. The system contains the largest-coverage grammar of English among implemented natural language parsers.