In this contribution, semi-blind channel estimation for joint communication and positioning (JCAP) is proposed. The concept is based on a unified signal architecture. Within this framework channel estimation is the core part, since it links communication and positioning algorithms. Thus, channel estimation serves twofold purposes: Firstly, it is inherently required for data detection and secondly it is employed to estimate positioning relevant information like the time of arrival (TOA). JCAP aims at exploiting synergetic effects. Employing semi-blind channel estimation is beneficial in this context, since it allows to assimilate the estimated communication data as virtual training.