The present study investigated whether and to what extent data-driven learning (DDL) can improve the lexico-grammatical use of abstract nouns in L2 writing. A corpus composed of 40 graded readers was compiled to make the learners do concordance learning activities, and 30 Turkish English as a Foreign Language (EFL) students at Gazi University School of Foreign Languages were assigned to a control group or an experimental group. At the prewriting stage, both the control group and the experimental group were given a list of ten abstract nouns and wrote stories without using dictionaries. Then, the errors they made while writing were underlined. While the experimental group was taught how to use a concordancing tool and studied on concordance lines from the corpus of graded readers to correct their errors, the control group just had dictionaries to consult and worked on their errors. Afterwards, both groups wrote their second stories using the same words given in the pre-test. The texts written in the pre-test and post-test were analysed and compared between groups. The results indicated that the experimental group, as compared with the control group, used a greater variety of collocational and colligational patterns and had fewer linguistic errors while using the abstract nouns. Finally, a questionnaire was administered to the experimental group and the results obtained from it showed that students were very positive about the use of DDL and concordance activities. Also, they were willing to use DDL activities in the future. (C) 2017 EJAL & the Authors. Published by Eurasian Journal of Applied Linguistics (EJAL).