With the growing complexity of railway control systems, it is required to preform runtime safety checks of system executions that go beyond conventional runtime monitoring of pre-programmed safety conditions. Runtime verification is a lightweight and rigorous formal method that dynamically analyses execution traces against some formal specifications. A challenge in applying this method in railway systems is defining a suitable monitoring specification language, i.e., a language that is expressive, of reasonable complexity, and easy to understand. In this paper, we propose parameterized modal live sequence charts (PMLSCs) by introducing the alphabet of the specification into charts to distinguish between silent events and unexpected events. We further investigate the expressiveness and complexity theories of the language. In particular, we prove that PMLSCs are closed under negation and the complexity of a subclass of PMLSCs is linear, which allows the language to be used to monitor a system online. Finally, we use PMLSCs to monitor an RBC system in the Chinese high-speed railway and evaluate the performance. The experimental results show that the PMLSC has high monitoring efficiency, and can reduce false alarm rate by introducing alphabets of charts.