The cerebellum sits at the back of the brain, contains more neurons than the entire cerebral cortex, and is traditionally understood as a motor structure — coordinating movement, timing, balance. When neuroscientists study language, they study the cortex: Broca's area, Wernicke's area, the temporal lobe, the frontal lobe. The cerebellum appears in language studies occasionally, noted and set aside, because the dominant framework locates language processing in cortical networks.
Published in Neuron, Evelina Fedorenko and colleagues at MIT used precision fMRI across 754 participants to systematically characterize cerebellar regions that respond to language. They found four. One region — spanning Crus I, Crus II, and lobule VIIb — is selectively responsive to language relative to diverse non-linguistic tasks. It activates during both comprehension and production, responds to linguistic difficulty, and processes both social and nonsocial sentences. It mirrors the cortical language network in its functional profile.
The structural insight is about what systematic measurement reveals that selective attention misses. The cerebellum was not absent from language studies. It appeared in data repeatedly. But it was not the region of interest, so it was treated as noise or secondary activation. The cortical framework determined which signals were investigated and which were noted without follow-up. Fedorenko's contribution is not discovering that the cerebellum responds to language — hints of this existed — but applying the same rigorous functional criteria used for cortical language areas to cerebellar areas for the first time. The response was always there. The measurement was not.
Language, like most cognitive functions studied since the 19th century, was mapped onto the structure where researchers expected to find it. The cerebellum was excluded not by evidence but by the framing of the question.