In this paper we propose a framework for the computational extraction of spatial and time characteristics of a single choreographic work. Computational frameworks can aid in revealing non-salient compositional structures in modern dance. The computational extraction of such features allows for the creation of interactive works where the movement and the digital feedback (graphics, sound etc) are integrally connected at deep level of structures. It also facilitates a better understanding of the choreographic process. There are two key contributions in this paper: (a) a systematic analysis of the observable and non-salient aspects of solo dance form, (b) computational analysis of spatio-temporal phrasing structures guided by critical understanding of observable form. Our analysis results are excellent indicating the presence of rich, latent spatio-temporal organization in specific semi-improvisatory modern dance works that may provide rich structural material for interactivity.