Relative Language Proficiency Modulates BOLD Signal Change when Bilinguals Perform Semantic Judgments Michael W. L. Chee, Nicholas Hon, Hwee Ling Lee, and Chun Siong Soon Cognitive Neuroscience Laboratory, Singapore General Hospital, 20 College Road, Singapore 169856, Singapore Received November 6, 2000 The effect of relative language proficiency on the spatial distribution and magnitude of BOLD signal change was evaluated by studying two groups of right- handed English–Mandarin bilingual participants with contrasting language proficiencies as they made se- mantic judgments with words and characters. Greater language proficiency corresponded to shorter re- sponse times and greater accuracy in the semantic judgment task. Within the left prefrontal and parietal regions, the change in BOLD signal was smaller in a participant’s more proficient language. The least pro- ficient performance was associated with right, in ad- dition to left, inferior frontal activation. The results highlight the importance of taking into consideration nature of task and relative language proficiency when drawing inferences from functional imaging studies of bilinguals. © 2001 Academic Press Key Words: Chinese characters; fMRI; semantic memory; bilinguals; language proficiency. INTRODUCTION Functional imaging studies have shown that blood flow changes occur in spatially congruent regions when proficient or relatively proficient bilinguals perform linguistic tasks in different languages (Chee et al., 1999a, 1999b, 2000; Illes et al., 1999; Klein et al., 1999; Perani et al., 1998). In these studies, the spatial pat- tern of activation appears to be similar across lan- guages irrespective of the surface features of the lan- guages compared. As there is evidence that proficient bilinguals can access concepts directly from second lan- guage (L2) without having to perform an internal translation through first language (L1) (Kroll and de- Groot, 1997), we suggested that common conceptual access is matched by the overlap of neuronal networks for processing L1 and L2 (Chee et al., 1999a). There is less agreement regarding less proficient bilinguals. Dehaene et al. (1997) reported inter- and intrahemispheric differences in spatial location when participants listened to sentences. Perani et al. (1996, 1998) reported a greater extent of activation for L1 when participants listened to sentences presented in L1 or L2, an effect that disappeared when participants were matched for proficiency in both languages. Fi- nally, marginally more prefrontal activation in the par- ticipants’ less proficient language has also been ob- served (Hernandez et al., 2000). We previously observed that Singaporeans (SGP) showed higher BOLD (blood oxygen level dependent) signal change when they performed a semantic judg- ment task in Mandarin, compared to when they did the same task in English (Chee et al., 2000). This finding differed from a previous study that drew from the same population but which did not show cross-language dif- ferences in activation while the participants evaluated sentence meaning (Chee et al., 1999a). In the study involving sentences, performance was closely matched across languages, whereas in the associative semantics experiment, performance was slower and less accurate in Mandarin, the participants’ less proficient language. A plausible explanation for the contrasting observa- tions is to attribute the differences in activation (and performance) to the relative proficiencies of L1 and L2. This effect of differential proficiency may have been evident in one task and not the other because of task- related-factors (e.g., overall difficulty of task). An alternative explanation is that there is a real difference in the processing demands between these languages, whereby Mandarin inherently requires more resources to process, and that this difference was somehow not revealed in previous experiments. Chinese (Mandarin and Chinese are used inter- changeably here) remains the only major language to use a purely logographic script. It has been proposed that this characteristic makes the concept represented by each character (or group of characters) relatively transparent to the reader (Wang, 1973). While it is true that the semantic radical in Chinese provides a clue to the meaning of the character, the extent to which the semantic radical is related to the concept actually de- noted by the complete character varies from vague to highly salient (Chen, 1999). Further, with common words in Modern Chinese, a good number of concepts NeuroImage 13, 1155–1163 (2001) doi:10.1006/nimg.2001.0781, available online at http://www.idealibrary.com on 1155 1053-8119/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved.