Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Sentence-level meaning and compositionality in a left fronto-temporal network
Poster E47 in Poster Session E, Thursday, October 26, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
Daniel Sharoh1,2, Lenno Ruijters1, Kirsten Weber1,2, David G. Norris1,3,4, Peter Hagoort1,2; 1Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, Netherlands, 2Max Planck Institute for Psycholinguistics, Nijmegen, Netherlands, 3Erwin L. Hahn Institute for Magnetic Resonance Imaging, Essen, Germany, 4University of Twente, Faculty Science and Technology, Dept. of Clinical Neurophysiology (CNPH), Netherlands
Neurobiological systems which process syntactic and semantic information are relatively well studied compared to those which support the integration of information across these domains. This type of integration is often critical for computing the sentence-level meaning of a given sentence, and is therefore of importance to neurobiological models of language which extend beyond single words. The known role of the left inferior frontal cortex (LIFC) in integrating linguistic and non-linguistic information (Lai et al., 2015; Özyürek et al., 2007; Willems et al., 2007) —as well in supporting combinatorial operations within linguistic domains (Hagoort, 2014)—suggests that it might also be involved in integrating information across linguistic domains. Based on this prediction, we conducted a 3T FMRI priming experiment (N=19) to investigate the role of the LIFC and the left perisylvian cortex in supporting cross-domain integration, which results in the compositional outcomes that give rise to sentence-level meaning. METHODS: In the experiment, participants were asked to read well-formed sentences and scrambled versions of these sentences, in which the word order did not form a proper sentence. In the critical condition, the prime was followed by a target word that was related to the sentence-level meaning of the prime, and was expected to reactivate this meaning more so than the other experimental conditions. There were four priming conditions in the experiment: (1) target primed by the sentence-level meaning of the sentence prime, (2) target primed by a repeated word embedded in the sentence prime, (3) target primed by a repeated word in the scrambled sentence prime, (4) target unrelated to any of the words in the scrambled sentence prime. Condition 1 was the critical condition, and the other conditions were included to account for the effect of sentence reading and for priming effects related to the reactivation of individual words and their local context. Data were analyzed both in native and MNI normalized space. An FROI analysis was performed in native space to identify the effects of the sentence meaning prime in regions throughout left perisylvian cortex. In addition, a task-dependent connectivity analysis was performed to assess the interaction between LIFC and left temporal regions known to be involved in sentence processing. RESULTS: The findings of this analysis demonstrate that a portion of the left mid temporal cortex supported syntactico-semantic integration at the level of sentence meaning in a network with the left inferior frontal cortex. This region was distinct from a more posterior left temporal region which was likely involved in lexical level processes and the integration of individual words into the sentence context. We also demonstrate that connectivity between the LIFC and left posterior temporal cortex was not modulated by the difference between sentence-level meaning and words read in sentence context, despite their previously observed role in combinatorial operations (Snijders et al., 2009; 2010). In light of these results, we propose the left mid temporal cortex/LIFC network as a target for future research on the neurobiology of the syntax-semantic interface.
Topic Areas: Syntax and Combinatorial Semantics,