The study aimed at identifying and describing the phonological differences between Somor and Aror sub-dialects of the Tugen dialect of the Kalenjin language. The Kalenjin language is characterised by eight dialects of which Tugen is one. These dialects can be mapped according to the geographical location of speakers. Of particular interest is the fact that there exists a dialect continuum, where neighbouring groups understand each other well, while mutual intelligibility decreases gradually to near-zero depending on the geographical distance. Even within given dialects, variations exist to significant levels. The study was guided by Natural Generative Phonology. Stratified and random sampling procedures were used to get samples of Somor and Aror speakers from the population of those who practice in the selected domains in Torongo and Kapuskei locations of Baringo County. The data for the study was a Swadesh list of one hundred and fifty words and fifty sentences. These were drawn from the fields of education, domestic life, religion, health and administration. Data was collected by use of language performance test, which was recorded, on an audiotape. These words were written in gloss and transcribed using the IPA symbols. This was in preparation for the phonological analysis, which was done by using Natural Generative Phonology and Descriptive Linguistics. This study will be an addition to the knowledge in the area of theoretical linguistics of Nilotic languages and Kenyan languages in general.