The research paper investigates variation in the prevalence and distribution of ‘voice’ in the spoken (presentations/interviews) and written (essays/midterm exam scripts) academic registers of BS English students of Khyber Pakhtunkhwa province and the relation of this variation to the situational circumstances of the concerned registers. The learner corpus was constructed. The data was collected from BS English students of the twelve public sector universities of Khyber Pakhtunkhwa. All the texts were transcribed, coded, and then analysed through the computer software AntConc (3.5.8). The present study is unique in the sense that it has conducted a comparative analysis of the various spoken and written sub-registers of English in the academic non-native environment. This study has great implications for researchers in the field of register analysis, material design, and language teaching.