Pas de résumé français / In this thesis, we investigated troubles of sentence processing in Huntington’s disease (HD) at earlystage, which represents a model of damage mainly confined to the striatum. The role of striatalstructures in sentence processing is agreed upon, but its nature is still controversial. Most studieshave reported a role of the striatum for complex sentences or more controlled processes withinsentence processing, but the interpretation of this pattern differs according to two main views.Some authors have proposed that striatal structures have a linguistic function restricted to somesub-processes of sentence processing, while others claim that the deficits detected depend on themodulation that executive function exerts on language and sentence processing. Here, we aimed atfilling the gap between these approaches by using a psycholinguistic perspective to investigate onthe one hand the role of executive functions, in particular, working memory, in sentenceprocessing, and on the other hand the nature of the linguistic discrepancies reported in associationto striatal lesions. We thus built experimental paradigms that allow dissociating fine-grainedlinguistic variations in healthy subjects, and transferred them to HD patients. The profile ofimpairment detected in our experiments showed that working memory and other sources ofcomplexity can interfere with sentence processing by decreasing accuracy, but an impairment ofspecific syntactic processes occurs when working memory is controlled for. The pattern of the finegrainedsyntactic impairment detected is consistent with a dissociation between more frequent/lesscontrolled (default) and less frequent/more controlled (non-default) procedures in sentenceprocessing. Additionally, we detected that this deficit occurs despite the fact that HD patients arestill able to process syntactic information, suggesting that striatal structures spare syntacticrepresentations while they are involved in correctly applying syntactic procedures in non-defaultcontexts. We propose that this pattern is explained by a role of striatal structures in selectingbetween competing alternatives during sentence processing, which results in an inability to adapt tothe sentence context for non-default procedures. This parallels the role of striatal structures for selecting between competing alternatives in order to adapt to the changing environment, as reportedin motor control and in other domains of cognition. Although the domain specificity of striatalinvolvement in language cannot be demonstrated, it is highly compatible with the results obtainedin this thesis. Hence, linguistic functions might be modulated by distinct cortico-striatal circuits: onthe one hand by selecting linguistic representations as a function of the context, and on the otherhand, by modulating performance in language through executive functions. The frameworkemerging from this work thus helps conciliating apparently incongruent findings reported in theliterature. Yet, future research should better characterize anatomo-functional correlates of thisproposal