In real-world scenarios, robots often interact with humans across various levels of abstraction, posing research challenges like task allocation, human-robot joint actions, and social navigation. To achieve seamless interactions and make robots more acceptable, they must adapt their behavior based on context, e.g., to navigate social contexts effectively. This paper proposes an approach integrating task and motion planning to contextualize robot behaviors for social navigation. It leverages the contextual knowledge of a deliberative task planner to adjust the robot navigation dynamically, improving human-robot interaction. The proposed framework can handle varied sequences of social navigation tasks, and adapt navigation behaviors to the interacting features of involved human users. We validate this approach through a simulated hospital scenario and show an in-laboratory deployment on a real robot.