摘要：Data-Driven Learning (DDL) approach is a pedagogic application of corpus linguistics in classroom. This paper aims at investigating the impact of DDL instruction (an integration of both paper-based (scaffolded) concordance materials or "soft" DDL and "hard" DDL (independent online searching) on the production of colligations of prepositions among law undergraduates at a university in Malaysia. In this experimental study design, data were collected from the pretest and posttest conducted before and after a seven week-course. The 40 respondents (third semester law undergraduates) involved in this study were divided equally. Twenty students (5 males and 15 females) were placed in the experimental group and treated with the module (concordance printouts of the colligational patterns) and the DDL approach—an inductive or a process approach. Meanwhile, the other 20 students (6 males and 14 females) were placed in the control group, and they were treated with the non-DDL module and taught deductively following the traditional teaching approach—a product approach. The data collected were then analysed using the Mann-Whitney U test (inferential statistics). The statistical analysis showed that the students in the DDL group performed significantly better than the students in the comparison group in the gap-filling task (U = 120.5, Z = –2.209, p = 0.027, significant at p < 0.05) and in the error- identification and correction task (U = 124.5, Z = –2.070, p = 0.038, significant at p < 0.05). However, no significant difference was found in the single-sentence writing (construction) task (U = 135.5, Z = –1.770, p = 0.077). This study recommends explicit teaching of colligations of prepositions via DDL approach (an integration of paper-based and online searching), as opposed to the non-DDL approach to law students
