Published online by Cambridge University Press: 28 November 2008
It has been repeatedly suggested in the literature on generative phonology that the number of features which appear in phonological rules be used in an evaluation measure for alternative descriptions of the phonology of a language; thus Halle states (1962: 55): ‘Given two alternative descriptions of a particular body of data, the description containing fewer… symbols will be regarded as simpler and will, therefore, be preferred over the other.’ The purpose of the present paper is to examine critically the application of the proposed feature-counting evaluation measure to alternative analyses of a very restricted body of data in the phonology of Turkish. The data considered are, as far as I can see, self-contained in that the analyses to be discussed do not have any implications elsewhere in the system of phonological rules; the adequacy of the decision based on the evaluation measure in question can therefore be fully assessed in connexion with the minor problem we shall be examining.