Data collected during a phonemic fluency task (or 'FAS test'), a standard component of neuropsychological batteries for assessment of cognitive deficits, may be language-dependent and may differ depending on second-language proficiency. The unique orthographic/phonological system of the task language, and the reported cognitive advantages inherent to bilinguals, may each influence the task's neural correlates. However, language background is not currently assessed in most studies testing phonemic fluency. Here, we used 52-channel functional near-infrared spectroscopy in college-aged native-Japanese subjects to examine functional changes in oxygenated hemoglobin elicited during a phonemic fluency task performed in Japanese and in English. We found activity differences that were related to task language and second-language proficiency. Besides loci activated in the Japanese test, bilateral precentral channels were specifically recruited in the English test. Furthermore, the higher-proficiency group showed almost no increase in oxygenated hemoglobin in either language context, whereas participants with lower proficiency showed widespread increases for both contexts. We interpret precentral increases as the consequence of additional articulatory resource recruitment in a second-language context. As for the lack of such variation in the higher-proficiency group, it may reflect an advantage in nonverbal executive control in this group. Together, our results point to language background and proficiency as confounding variables in neuroimaging studies of phonemic fluency and that the adequacy of such measures in populations with varying language backgrounds needs to be considered in future studies.