This paper argues for the integration of the lexical approach with a data-driven corpus-based methodology in English teaching for technical students, particularly students of Engineering. It presents the findings of the author's computer-aided research, which aimed to establish a frequency-based corpus of student engineering lexis. The Student Engineering English Corpus (SEEC), reported here, contains nearly 2,000,000 running words reduced to 1200 word families or 9000 word-types encountered in engineering textbooks that are compulsory for all engineering students, regardless of their fields of specialization. The most immediate implication arising from this research is that sub-technical vocabulary as well as Academic English should be given more attention in the ESP classroom. The paper illustrates some sample data-driven instructional activities consistent with the lexical approach, in order to help students acquire the so-called "language prefabs", or formulaic multi-word units/collocations, for technical and non-technical uses. The integration of the lexical approach with a corpus linguistic methodology can enrich the learners' language experience and raise their language awareness, bringing out the researcher in them.