As a common sense, terahertz (THz) communications require the assistance of high-accuracy sensing to deal with the mobility challenge. Existing works tried to measure the azimuth angle of signals or targets to decease beam pointing errors, while it would fail in scenarios with high mobility without angular velocity information of targets. In this work, a novel THz sensing-assisted communication (TSAC) framework where both angle and angular velocity information extracted from sensing echoes can be utilized for beam alignment is introduced. Capitalizing on the non-uniform amplitude nature of THz antennas in angular domain, the four-dimensional target state parameters including distance, angle, radial and angular velocity are estimated. In addition, to exhibit superiority of the TSAC framework, the beam alignment variances applicable to three assistance cases are derived. Hence, the performance bounds of sensing and communications under Gaussian beam model are concretely obtained in form of Cramér-Rao Bounds (CRBs) and Shannon Capacities. Finally, numerical results in different bandwidth factors, angular velocities and beam widths are performed to validate the framework, proving the benefits of angular velocity estimation in TSAC scenarios.