Code switching (the process of switching from one language to another during a conversation) is a common phenomenon in multilingual environments. Where a minority and dominant language coincide, code switching from the minority language to the dominant language can become particularly frequent. We analyse one such scenario: Sepedi spoken in South Africa, where English is the dominant language; and determine the frequency and mechanisms of code switching through the analysis of radio broadcasts. We also perform an initial acoustic analysis to determine the impact of such code switching on speech recognition performance. We nd that the frequency of code switching is unexpectedly high, and that the continuum of code switching (from unmodi ed embedded words to loan words absorbed in the matrix language) makes this a particularly challenging task for speech recognition systems.
Reference:
Modipa, T.I, Davel, M.H and De Wet, F. 2013. Implications of Sepedi/English code switching for ASR systems. In: Conference Proceedings of the 24th Annual Symposium of the Pattern Recognition Association of South Africa, Johannesburg, South Africa, 3 December 2013
Modipa, T., Davel, M., & De Wet, F. (2013). Implications of Sepedi/English code switching for ASR systems. PRASA 2013 Proceedings. http://hdl.handle.net/10204/7287
Modipa, TI, MH Davel, and Febe De Wet. "Implications of Sepedi/English code switching for ASR systems." (2013): http://hdl.handle.net/10204/7287
Modipa T, Davel M, De Wet F, Implications of Sepedi/English code switching for ASR systems; PRASA 2013 Proceedings; 2013. http://hdl.handle.net/10204/7287 .
Conference Proceedings of the 24th Annual Symposium of the Pattern Recognition Association of South Africa, Johannesburg, South Africa, 3 December 2013