Introduction

In this chapter I examine the two principal segmentally based theories of phonological representation, phonemic and transformational phonology. These were the first ‘class’ of phonological theories which historically arose from segmental phonetic theory. After a brief historical resumé, I shall discuss the theory of phonemic phonology, the representation of distinctive segmental phonological units, and phonological representations made up of those units. In section 3.4, I describe transformational-generative phonology, a derivational theory of the mapping from lexical (phonological) to surface (phonetic) segmental representations. I illustrate some formal problems of transformational grammars with a transformational phonological analysis of three kinds of allophony in Japanese. Despite the wide use of and apparent empirical support for the rewrite-rule notation and the primitive operations of insertion, deletion and replacement in generative phonology, I argue in section 3.5 that the grammar formalism employed in transformational-generative phonology is excessively unconstrained, being equivalent to unrestricted rewriting systems. I also argue that the constraints on transformational phonology which have been proposed cannot be relied upon to overcome the problems of excessive power.