This study set out to investigate the roles of the two different types of structured input activities (SIA) in the framework of Processing Instruction (i.e. PI, a type of input-based grammar pedagogy package). It addresses some weaknesses of previous PI studies and begins to address several new issues in the PI research agenda. A classroom-based quasi-experiment was carried out to investigate the effectiveness of different types of input activities within PI: Referential and affective activities. Taiwanese primary school students were allocated into instruction and control groups. A range of elicitation tests were employed to measure the impact of the interventions, including a grammaticality judgment test (GJT) with a time constraint, a gap-fill written test, a picture-based narration oral test, a structured conversation and a short written receptive vocabulary test.