In this study I explored the possible emergence of a rural/urban contrast in the vocabulary of Xhosa. In order to focus the study I used 15 Xhosa words that appeared to be undergoing semantic shift and researched the meanings ascribed to these words with Xhosa-speakers in both rural and urban areas. Using a sociolinguistic theoretical framework I investigated the connection between socio-demographic data (as well as language attitudes held by speakers) and the influence of these factors on semantic choice. I used a qualitative methodological design, with purposeful sampling in these areas. The reason for selecting these areas was in order to incorporate both urban and rural styles of speech. Data was collected by means of open ended questionnaires and structured interview questions with participants across a wide age-range. This data was carefully analyzed according to how speakers use these terms during their conversations and in writing, and was then recorded, summarized and interpreted by means of descriptive statistics. The results of this study suggest that a possible urban/rural contrast is emerging in the lexicon of Xhosa speakers, with young, urban speakers being responsible for more innovative semantics than their counterparts in rural areas. The study also examined motivations for the semantic shift in the selected terms – sociocultural and emotionally marked motivations would appear to dominate with speakers choosing new meanings to suit their altered cultural environments and a new openness to talking about casual sex. The dominant semantic processes involved metaphorization, metonomy, amelioration, pejoration and bleaching, with most of the selected words becoming more polysemous. From these results recommendations are made for terminology development, dictionary revision, mother-tongue education and the creation of literature in Xhosa using the vocabulary of the youth. I conclude that further research on semantic shift in Xhosa is necessary in order to provide these sectors with reliable and useful data.