The assimilation hypothesis argues that second language learning recruits the brain network for processing the native language, whereas the accommodation hypothesis argues that learning a second language recruits brain structures not involved in native language processing. This study tested these hypotheses by examining brain activation of a group of native Chinese speakers, who were late bilinguals with varying levels of proficiency in English, when they performed a rhyming judgment to visually presented English word pairs (CE group) during fMRI. Assimilation was examined by comparing the CE group to native Chinese speakers performing the rhyming task in Chinese (CC group), and accommodation was examined by comparing the CE group to native English speakers performing the rhyming task in English (EE group). The CE group was very similar in activation to the CC group, supporting the assimilation hypothesis. Additional support for the assimilation hypothesis was the finding that higher proficiency in the CE group was related to increased activation in the Chinese network (as defined by the CC > EE), including the left middle frontal gyrus, the right inferior parietal lobule, and the right precuneus, and decreased activation in the English network (as defined by the EE > CC), including the left inferior frontal gyrus and the left inferior temporal gyrus. Although most of the results support assimilation, there was some evidence for accommodation as the CE group showed less activation in the Chinese network including the right middle occipital gyrus, which has been argued to be involved in holistic visuospatial processing of Chinese characters.