This paper employs data from elementary school children in a bilingual classroom to examine the applicability of two models for bilingual speech: the Markedness Model (Myers-Scotton 1993; Myers-Scotton & Bolonyai 2001) and Sequential Approach (Auer 1988, 1995; Li Wei 1988). The majority of the switches can be explained within either model, but each approach stresses different aspects of bilingual discourse. While the Markedness Model is preferred because it sheds light on social identities, in cases of codeswitching as the unmarked choice, it cannot account for individual switches. In such interactions, it is shown that the additional application of the Sequential Approach can be used to shed light on the conversational structure of bilingual discourse.