Grammatical voice is an important element of computer science discourse as an effective rhetorical means of establishing disciplinary membership and describing the procedures and processes in the research methodologies of a rapidly expanding, cosmopolitan discipline. This particular relationship between verbs and their arguments has proved especially challenging for Asian students as a result of not only L2 structural complexity but also L1 conceptual interference. The question of whether to include voice in an ESP program in Japanese tertiary contexts may be further complicated by both lack of available classroom time and falling English proficiency levels of incoming students. In this paper, we describe a pilot project aimed at teaching grammatical voice to computer science students in a Japanese university setting. The instruction comprised a three-week concept based unit based upon a sociocultural understanding of language development and included a grammaticality judgment task as part of a pretest-posttest quasi-experimental design. The findings confirm the difficulty of teaching the various facets of voice to learners
with low English proficiency, yet some significant gains were also made. Close analysis of the data suggests that coordinated instruction in the metaphorical underpinnings of different aspects of grammatical voice may better inform the teaching of voice in the English for Computer Science writing syllabus.