This paper investigates the effect of multiple cultures and languages on the bilingual utterances of Libyan children who live in the United States and who have acquired English after they arrived there at ages from 3 to 5. Data analysis is based on the Markedness Model (Myers-Scotton, 1993) in order to determine which language is the unmarked code and which language is the marked one. According to Myers-Scotton, the unmarked code is mostly the native language, which is also supposed to be the dominant language. The bilingual utterances in this study are analyzed in terms of subjects' responses to the interviewer and the culture-specific topic under discussion; some of the responses are made in a code different from the one in which the question was asked; others are culture-related. The results show that it is the dominant language (not necessarily the native language) that is most likely the unmarked code, and the less dominant language is the marked code which is chosen more consciously. (Contains 5 tables and 2 figures.)