A TE does not read program code. Instead, a TE starts with the systems specifications produced by SAs. Then, a TE derive test plans to check whether an information system works right. A test plan consists many steps. In each step, there is a stimulus (what to do to the system) and an expected response (how the system should react). Although a test plan sounds easy, it is extremely difficult for large and complex systems.
If an information system fails a test, the failure is reported to SAs and developers. Note that a failure means that the system does not behave as specified. This can be a misinterpretation of systems specifications, or incorrect implementation.