Does nonsense make sense? A springboard to studying dynamic reconfiguration of large-scale networks during semantic...

Along with semantic content, prosody, which includes intonation, stress, and rhythm, plays an important role in speech comprehension by conveying paralinguistic information. While functional neuroimaging data yields valuable insights into the involvement of language-related grey matter structures in semantic and prosodic processing, the dynamic organization of large-scale functional networks supporting ecologically valid, continuous speech perception remains poorly understood. By selectively disrupting the semantic and/or prosodic structure of complex, continuous speech narratives, we investigated how semantic coherence and intonation modulate interactions between language network and domain-general systems. Our factorial fMRI design revealed involvement of classical left-lateralized language network in the processing of semantic content and bilateral superior temporal areas in the processing of intonation. Scrambled speech engaged the salience network and executive control regions, reflecting effortful meaning extraction. Psychophysiological interaction (PPI) analysis revealed a functional dissociation within the inferior frontal gyrus (IFG). Semantic coherence strengthened connectivity within left-hemispheric language network, including both IFG subregions, pars opercularis (IFGop) and pars triangularis (IFGtri), posterior temporal cortex, and inferior parietal lobule (IPL), and between right IFG and left IPL. Intonation enhanced right IFGop coupling with left temporal cortex, while reducing left IFGop connectivity with the right IPL. Thus, bilateral IFGop demonstrated context-dependent flexibility across semantic and prosodic networks, whereas bilateral IFGtri was specifically engaged in semantic processing. These findings illustrate a dynamic trade-off, where the engagement of core language networks for structured speech is complemented by the recruitment of domain-general evolutionary older systems when processing demands increase due to degraded input.