In this contribution I will put forward a set of research questions which pertain to the differences between the normal and the pathological variation in speech. I will review the established clinical findings and discuss the new neurolinguistic models based on neuroimaging research, which attempt to explain the extent and the source of pathological variation. I will put the data in relation to the models and show which pathological patterns can be readily accounted for. It will become clear that some aspects of pathological variation in speech cannot be accounted for by the existing models. Interestingly, these aspects are those that keep pathological speech similar to normal speech. I will show that even the most disturbed forms of pathology in speech adhere to well-defined, general principles of phonological structure and phonetic implementation. These principles are what I will refer to as the “hard-wired” part of phonology, which emerges only when the pathological patterns are compared with the normal learning path. In the final part of this contribution I will attempt to identify the research questions and the research tools that might help us understand how the hard-wired phonology is implemented in the human brain.