A grammar of English monosyllables

Introduction

In chapters 3 and 5 it was shown that declarative grammars have the desirable property of being independent of any particular procedural implementation, so that a given grammar can be used with a variety of different parsing algorithms, or can just as easily be employed for generating sets of strings and structures. The grammar of English monosyllables which I will describe here has been implemented and thoroughly tested as part of a speech synthesis system (Coleman 1990b, 1992). The grammar described below was employed both for parsing input strings presented to the system, and for enumerating the set of input strings admitted by the grammar. In the normal operation of the speech synthesis program, the grammar is interpreted as a parsing program – a Prolog Definite Clause Grammar (DCG; Pereira and Warren 1980) – to determine the phonological structure of an input string. The structures determined in this way are then used to evaluate a compositional phonetic interpretation, without using rewrite rules. In order to ensure that the set of strings defined by the grammar was complete, the grammar was used generatively and its yield automatically tested against an on-line dictionary.

The phonotactic grammar of English will be discussed in two parts: syllable-level and word-level. The first part of this chapter (section 7.1) concentrates on the syllable-level subgrammar, and the second part, section 7.2, describes the word-level subgrammar.