Poster No.: 1054
Cortical Correlates of Affective vs. Linguistic Prosody: An fMRI Study
Mirella Dapretto, Ahmad Hariri, Mayim H. Bialik, Susan Y. Bookheimer

Brain Mapping Center, University of California, Los Angeles, CA, USA

Introduction. Neuropsychological (1) and neuroimaging (2) studies suggest that the right hemisphere (RH) is dominant for understanding the emotional aspects of speech (affective prosody). However, while some researchers have argued that the RH is dominant for all type of prosodic information (3), others (4) believe that processing linguistic prosody (i.e., the quality of speech that carries meaning information through syllable, word, or sentence level stress and intonation) may be subserved by the left hemisphere (LH). In the present study, we used fMRI to further investigate the neuroanatomic correlates of affective vs. linguistic prosody during an auditory sentence judgment task.
Activation Paradigm. A selective attention paradigm was used in this study where subjects listened to 8 pairs of sentences in each of 3 activation conditions. In all conditions, half of the sentences had a neutral intonation, whereas the remaining half differed in prosodic cues across conditions. In the Linguistic Prosody condition (LP) half the sentences had a rising pitch at the end of the sentence (turning a declarative sentence into a question); in the Affective Prosody condition (AP) half the sentences had either a sad (for half of the subjects) or an angry (for the other half of the subjects) intonation. In the Semantic Control condition (SC), half of the remaining sentences had a rising intonation as in LP, while the other half had either a sad or an angry emotional intonation as in AP. In the two prosody conditions the subjects had to judge whether the sentences in each pair sounded alike (half the pairs differed in intonation), regardless of whether they differed in literal meaning. In the control condition, the subjects had to determine if the sentences in each pair had the same meaning, regardless of their intonation (half the pairs had the same meaning).
Methods. Eight normal adult volunteers (4 males, 4 females), right-handed and native English speakers participated in this study. The activation conditions lasted 45 sec each and alternated with 4 rest periods of 22.5 sec each; for half the subjects the order of the 3 condition was AP-SC-LP, for the other half this order was reversed. Data were collected on a GE 3 Tesla scanner with ANMR upgrade for EPI. For each subject, 84 functional images were acquired over 17 axial/oblique slices using an asymmetric spin echo sequence (TR = 2.5; TE = 70 ms; offset = 180°; matrix size = 64 x 64; FOV = 20 cm; 4 mm slices/1 mm gap), together with a set of co-planar high resolution images (TR = 4000; TE = 46 ms; matrix size = 128 x 128; FOV = 20 cm). Each subject's data were realigned to correct for head motion, spatially normalized, and smoothed. Group analyses were then conducted with SPM96 using a delayed box-car reference function with a 6 sec delay to compensate for the lag in the hemodynamic response.
Results and Discussion. Not surprisingly, given the virtually identical stimuli and task demands across conditions, a substantially similar pattern of activation was observed for each activation condition compared to rest. For these contrasts, significant foci of activation were observed in primary auditory cortices (transverse and superior temporal gyri, BA 41 and 42), in Wernicke's (superior and middle temporal gyri, BA 21, 22, 39, and 40), and in Broca's area (inferior frontal gyrus, BA 44 and 45), and dorsolateral prefrontal cortex (middle frontal gyrus, BA 9), with the left hemisphere showing overall greater activation in terms of magnitude and spatial extent. The direct contrasts between activation conditions were also examined using the respective activation-rest contrasts as masks. Consistent with the results of a previous PET study (George et al, 1996), the Affective Prosody condition elicited selective activation in the inferior frontal gyrus (BA 45) in the RH (Fig. 1), though a second cluster of activation was also detected in the inferior parietal lobe in the dominant hemisphere (Wernicke's area, BA 40). In contrast, the Linguistic Prosody condition was associated with selective activation in the inferior frontal gyrus (BA 47) and in the superior temporal gyrus (BA 22) in the LH only (Fig. 2). Taken together, the present findings are consistent with the notion that prosodic cues conveying meaning vs. emotional information are in part subserved by different cortical networks showing distinct lateralization profiles.

1. Borod JC, et al. Neuropsychologia, 1992, 30:827-844.
2. George MS, et al. Arch Neurol, 1996, 53:665-670.
3. Shapiro BE, et al. Brain and Lang, 1985, 25:19-36.
4. Emmorey KD. Brain and Lang, 1987, 30:305-320.