English for the Computer: The SUSANNE Corpus and Analytic Scheme

S Sampson proposes a comprehensive annotation scheme, covering part-of-speech tagging, surface grammar, and issues of underlying “logical grammar,” with the hope that the scheme might provide the basis for a widely adopted standard.

Abstract

Computer processing of natural language is a burgeoning field, but until now there has been no agreement on a standardized classification of the diverse structural elements that occur in real-life language material. This book attempts to define a Linnaean taxonomy for the English language: an annotation scheme, the SUSANNE scheme, which yields a labelled constituency structure for any string of English, comprehensively identifying all of its surface and logical structural properties. The structure is specified with sufficient rigour that analysts working independently must produce identical annotations for a given example. The scheme is based on large sample of real-life use of British and American written and spoken English. The book also describes the SUSANNE electronic corpus of English which is annotated in accordance with the scheme. It is freely available as a research resource to anyone working at a computer conected to Internet, and since 1992 has come into widespread use in academic and commerical research environments on four continents.