Reduced functional activation in the right Crus II is coupled with increased cerebral connectivity in processing linguistic violations—A study based on the auditory language localizer task

Research has demonstrated that the posterolateral cerebellum, especially right Crus II, operate synergistically with the cerebral networks over many linguistic tasks. The cerebellum is also part of the extended language networks identified by language localizer tasks (see Fedorenko et al., 2024, Nature Reviews Neuroscience). Recent studies (e.g., LaBel & D’Mello, 2023, Current Opinion in Behavioral Sciences) have called for the incorporation of the cerebellum into the neuroscientific models of language processing, even though its precise role has remained under active investigation. The present study aimed to provide further evidence for characterizing the nature of the cerebellar involvement, by conducting both functional activation and generalized psychophysiological interaction (gPPI) analysis in an auditory language localizer task. 117 cognitively normal participants uniformly sampled from the age range 18-81 were recruited. A set of 120 syntactically, lexically, and semantically valid [Syn+ Lex+ Sem+] sentences, all 12-syllable long in Cantonese, were first composed. Three conditions were constructed by progressively removing semantic [Syn+Lex+Sem-], lexical [Syn+Lex-Sem-], and syntactic [Syn-Lex-Sem-] information, yielding four conditions in total: (1) Valid, (2) Semantic violation, (3) Jabberwocky, and (4) Randomized Syllables. A speech synthesizer was used to construct the auditory stimuli. Participants were asked to covertly repeat the final syllable of each sentence. T1-weighted structural scans were acquired using an MPRAGE sequence. Functional images were acquired with a multi-band EPI sequence (TR = 800 ms, TE = 37 ms, and flip angle = 52°), covering the whole brain with 72 interleaved axial slices (2×2×2 mm3). The functional data were analyzed using the Conn toolbox and custom-made Nilearn scripts. As expected, widespread functional activation was found between valid sentences and randomized syllables, with valid sentences eliciting stronger activation predominantly the left-hemispheric core language areas. In support for a substantial role of the cerebellum in linguistic processing, the right Crus II was more strongly activated by valid sentences than randomized syllables. Interestingly, the opposite pattern was found for left Crus I and II, as well as for a similarly broad activity in the right hemisphere (the lateral and medial superior frontal gyrus, middle frontal gyrus, anterior cingulate cortex, and superior temporal gyrus). GPPI analysis revealed that, during the processing of randomized syllables, there was increased functional connectivity between right Crus II and left superior and middle frontal gyrus, bilateral precentral gyrus, and left postcentral gyrus. Thus, although the right Crus II was less activated by randomized syllables, that was accompanied by increased connectivity with the frontoparietal regions. Convergent evidence was observed for semantic violation and Jabberwocky conditions. Taken together, such connectivity pattern, elicited by linguistic anomalies, likely represents the neural basis of an error-monitoring and error-correction mechanism that is mediated by the right Crus II region, in combination with the language control network. Further research is required to elaborate whether the right Crus II region is a major hub in this cerebrocerebellar network. This research has been supported by HKRGC-GRF 15601718 and 15609423.