Codeswitching, borrowing and mixing in a corpus of Xhosa English

Abstract

The paper analyses selected aspects of the codeswitching behaviour in a spoken corpus of the English of 326 people, all of them mother-tongue speakers of Xhosa (a local African language in South Africa), and all of whom would see themselves as Xhosa/English bilinguals. The corpus comprises approximately 550,000 transcribed words of spontaneous, relaxed, oral discourse in English between pairs of Xhosaspeaking interlocutors, discussing a wide range of topics. While the usual pattern in bilingual speech is to use the L1 as matrix language and the L2 as embedded language, in this corpus the opposite is the case, as interlocutors were interviewed in English (the L2). The corpus therefore offers a ‘mirror image’, in a sense, of normal codeswitching behaviour. Using Wordsmith (a concordancer programme), all incidences of codeswitching into Xhosa during these conversations were identified and analysed in an effort to reveal underlying patterns. Examination of the amount and nature of codeswitching in the corpus promised to throw some light on the extent to which participants are genuinely bilingual, in terms of their ability to converse comfortably in English.