We present an automated method of identifying and representing non-emotional facial expressivity in video data. A benchmark dataset is created using the framework of an existing clinical test of upper and lower face movement, and initial findings regarding automated quantification of facial motion intensity are discussed. We describe a new set of features which combine tracked interest point statistics within a temporal window, and explore the effectiveness of those features as methods of quantifying changes in non-emotional facial expressivity of movement in the upper part of the face. We aim to develop this approach as a protocol which could inform clinical diagnosis and evaluation of treatment efficacy of a number of neurological conditions including Parkinson’s Disease.