Investigating language processing of syntax and prosody via dynamic causal modelling (DCM) on MEG data

Abstract

Event Abstract Back to Event Investigating language processing of syntax and prosody via dynamic causal modelling (DCM) on MEG data Burkhard Maess1*, Olivier David2, Korinna Eckstein1 and Angela D. Friederici1 1 Max Planck Institute for Human Cognitive and Brain Sciences, Germany 2 INSERM U836, Grenoble Institut des Neuroscience, France Our study investigated the interplay of syntactic and prosodic information processing during language comprehension. For this purpose, we conducted a MEG-study using spoken sentences which were either correct, syntactically incorrect, prosodically incorrect or both [1]. Averaged MEG data were subjected to DCM analysis. DCM priors for the activity centres were taken from previous functional imaging studies. We included six cortical regions in the analysis: Heschl’s gyrus (HG), the (anterior part of the) superior temporal cortex (STG) and the frontal opercular cortex (FOP) of both hemispheres. Recent studies have indicated the importance of the basal ganglia and the thalamus for language processing. Therefore, we also tested the influence of such a deep but magnetically silent source. According to current knowledge on the structural connectivity between these regions, we defined the following functional connections in all models: Within each hemisphere, HG connects to STG and STG to FOP. Fifteen different models, which differed with respect to their connections between the hemispheres and their connections to the thalamus, were tested. Model comparison supported most models which had connections to the thalamus from all six cortical regions and temporal as well as interhemispheric connections via the corpsus callosum between STG and FOP. Hence DCM modelling stresses the role of the thalamo-cortical loop as a modulator of the STG and FOP crosstalks within as well as between the hemispheres. For syntactic violations, a significant increase in connectivity strength was observed mostly for the left hemispheric connections to the thalamus. Prosodic incongruencies strengthened STG-Thalamus connections while also weakening the direct interhemispheric connection. In conclusion, using DCM, we were able to model the spatio-temporal evolution of brain activity during language processing. Furthermore, the crucial role of the thalamus within the language network was demonstrated.