Despite their large advances, especially for consumer applications, automatic speech technologies still encounter very huge difficulties when they are exposed to dysarthric speech. However, they were presented very early as potential solutions to provide objective tools to deal with speech disorders in order to help clinicians in their clinical practice and patients in their everyday lives. In order to understand the difficulties encountered by automatic speech processing, this paper investigates the reliability of a simple automatic phone alignment when dealing with dysarthric speech. Notably, the corpus used involves French read-speech recordings produced by patients suffering from four different pathologies, exhibiting three different types of dysarthria. The observations of the segmentation outputs yielded by the automatic tool (compared with a manual segmentation) according to the pathologies, the type of dysarthria and different phonetic categories reveal a very large heterogeneity of behavior between pathologies, and within the same pathology.