Abstract
This paper has three, related goals. Its main goal is to provide quantitative evidence supporting the claim that in bilingual conversation the unmarked choice can be identified via a frequency-based criterion(cf. Myers-Scotton,1993b). Accordingly, frequency also identifies the marked choice. Second, it shows that not all participants in the same conversation necessarily have the same unmarked choice. Data come from a Malawian family temporarily living in the United States. Both parents and the children engage in codeswitching, but how the two languages are employed and their frequency within the overall codeswitching pattern shows that for the parents, English is marked, while for the children Chicheŵa is marked. Finally, the paper reinforces the argument that people are rational actors in the sense that they perceive selecting one way of speaking over another as a means to optimize their outcomes in interpersonal relationships(cf. Myers-Scotton, 1998; Myers-Scotton & Bolonyai, 2001). When speakers are bilingual and codeswitching is a component of the community, the types of code choices they make in codeswitching conversations become a means of achieving such goals. Specifically, marked choices become self-presentations.