This exploratory study describes a framework for data-driven learning (DDL), in General (non-major) English university classes, in which learners "construct" linguistic corpora instead of merely "consulting" them. Prior related work has addressed the needs of language specialists, in particular trainee translators who are learning how to compile glossaries, rather than non-major students of English. It is argued in this article that the process of creating a corpus inculcates a sense of ownership in the learner and therefore has a motivational impetus. This is especially true, it is claimed here, when the topic of the corpus is of personal interest to the learner, or coincides with their major field of study. Learners may pursue language study for only a short period of their university career, but once the corpus is constructed, some students may be sufficiently motivated to consult it and add to it when needed. Moreover, the process of compiling the corpus may lead to the acquisition of not only language but also useful transferable skills, including information technology and problem-solving competencies. This study presents some of the motivational issues surrounding DDL in Asia and suggests corpus construction as a solution. Previous research on corpus construction by learners is reviewed. In the experiment which forms the core of this study, 90 freshmen compiled and analyzed corpora as part of a General English course in Taiwan. Of these, 19 students completed final projects based on corpora they had compiled. Their findings--and reactions to the use of corpora compilation as a language learning tool--are reported in a qualitative data analysis. (Contains 4 tables and 2 figures.)