Tan Yingying, Zhang Ran, Ye Zheng, Zhou Xiaolin
The basal ganglia (BG) are a group of subcortical nuclei that play critical roles in motor control, reinforcement learning, and language processing. Anatomically, the BG include the striatum, globus pallidus, subthalamic nucleus, and substantia nigra, which are closely connected to the cerebral cortex. Traditionally, the BG have been implicated primarily in motor functions. However, emerging evidence has shown that they also play important roles in language processing. Patients with BG dysfunction, such as those with Parkinson's disease or Huntington's disease, exhibit not only motor impairments but also a range of language disorders. These findings challenge traditional Broca-Wernicke-Geschwind language model, suggesting that both cortical and subcortical structures, particularly the BG, are essential for language function.
While earlier research has primarily focused on the role of the BG in speech production, recent studies have expanded the scope to encompass language comprehension. Although language comprehension and production share certain neural mechanisms, they also involve distinct processes. It remains unclear, however, whether the different BG nuclei contribute differently to various aspects of language comprehension, such as semantics, syntax, phonology, and pragmatics. This review summarizes research using methods from neuropsychology, neuroimaging, and psychopharmacology to address these questions.
Our review indicates that the dorsal striatum, composed of the caudate and putamen, is crucial for regulating almost all aspects of language comprehension. During both semantic and syntactic processing, the dorsal striatum is involved in monitoring and modulating information selection, activating goal-relevant information while inhibiting irrelevant or less-preferred ones. Neuroimaging studies reveal an anterior-posterior gradient within the dorsomedial striatum, with more anterior regions supporting complex syntactic processing. Moreover, the dorsal striatum is involved in pragmatic processing, as it coactivates with the frontal-temporal network to generate context-appropriate meanings. Studies also suggest that the putamen contributes uniquely to phonological processing. In contrast, the globus pallidus and subthalamic nucleus, which are commonly targeted in deep brain stimulation (DBS) for medical treatment, are primarily linked to speech production. Although some studies suggest their involvement in semantic and syntactic processing, others have failed to observe activation in these areas. Further research is necessary to clarify the precise role of these nuclei in language processing.
Beyond the BG nuclei themselves, the catecholaminergic (CA) system—particularly dopamine (DA) and norepinephrine (NA)—plays a crucial role in language comprehension via the fronto-striatal pathway. Closely interconnected with the basal ganglia, this system not only modulates motor function but also supports higher-order cognitive processes, including multiple aspects of language. The role of CAs in semantic processing has gained considerable attention. Patients with BG dysfunctions often exhibit deficits in semantic, syntactic, and pragmatic processing. Recent pharmacological studies in healthy individuals have shown that CA stimulants (e.g., levodopa and methylphenidate) causally enhance the semantic and syntactic unification, even when language processing per se is goal-irrelevant. These findings demonstrate that higher CA levels may further amplify the importance of language processing through modulating fronto-striatal connectivity.
A central question in current research is whether the BG support language comprehension through mechanisms that are language-specific or domain-general mechanisms. The domain-general view, supported by evidence from neuroimaging, neuropsychological, and bilingual-switching studies, proposes that the BG—particularly the caudate nucleus and putamen—modulate cognitive control processes such as selection, monitoring, and resource allocation across both linguistic and non-linguistic tasks. In contrast, the language-specific view argues that certain BG-frontal circuits are specialized for language, with some neuropsychological and fMRI studies showing comprehension deficits or heightened fronto-striatal sensitivity in language tasks independent of general executive dysfunction. While current findings largely favor a domain-general regulatory role, the BG may exert finer-grained, potentially specialized control in specific linguistic domains, such as syntactic processing. Resolving this issue is critical for addressing a longstanding debate in psycholinguistics: whether language processing relies solely on shared neural resources or also engages dedicated neural mechanisms.
In summary, this review underscores the distinctive role of the BG in language comprehension, highlighting functional specializations among its nuclei. Several critical questions, however, remain unanswered. Future research should move beyond isolated activation patterns to systematically map the division of labor and coordination among BG nuclei, as well as their interactions with cortical language networks, using advanced techniques such as high-resolution diffusion imaging, laminar fMRI, intracranial recordings, and psychopharmacology. A central priority is to clarify whether BG regulatory functions are language-specific or domain-general by directly comparing linguistic and non-linguistic tasks with temporally precise (MEG/EEG) and spatially precise (ultra-high-field fMRI) measures. The catecholaminergic system—particularly dopamine and norepinephrine—also warrants focused investigation to disentangle their distinct and potentially non-linear contributions, integrating pharmacological, genetic, and MR spectroscopy approaches to assess neurotransmitter concentrations, receptor distributions, and connectivity effects. Finally, the role of neural plasticity, especially the bilingualism-induced structural and functional adaptations, should be incorporated into dynamic, developmentally informed models of BG-language interaction. These models should be supported by longitudinal, multimodal imaging to link language experience with functional network reorganization.
