Levels of representation
In this chapter, we consider the question of how many phonological representations a word has. So far we have postulated two levels of representation, an underlying one and a surface one, a position that was motivated in Chapter 5. It can be argued, however, that this two-level model does not account for the intuitions of native speakers about the pronunciation of the words of their language. In brief, the underlying representation would seem to be too abstract, while the surface representation appears to be too detailed (cf. Schane 1971). This unsatisfactory state of affairs has come to an end with the advent of Lexical Phonology, which theory postulates an intermediate level of representation known as the lexical representation. Not only does it corres pond to native speaker intuitions, it will also be shown to have a number of interesting properties. After dealing with the lexical representation, we turn our attention to the relation between the surface representation and the physical pronunciation of the words. We will see that languages may differ in the way they realize identical surface representations. Such language-specific realizations of phonological elements are accounted for by rules of phonetic implementation.