Yeh, Hsin-Jou

Data-Driven Learning (DDL) is a corpus-based instructional approach in which learners acquire language knowledge through observing and analyzing large amounts of authentic language data. This study first reviews a teaching experiment conducted by the author in 2024 with 17 CSL learners, outlining its instructional design, implementation procedures, and major findings. The results of the experiment indicate that DDL can effectively enhance high-intermediate learners’ ability to distinguish near-synonyms. Building on these findings, the present study further examines how DDL can be adapted and implemented in an authentic CFL classroom under non-experimental conditions. In a high-intermediate oral course offered during the current semester, the author employed Sketch Engine as the corpus tool and incorporated multiple in-class corpus-based activities, each lasting 15 to 30 minutes, as part of lexical instruction. These activities guided 7 German learners to observe collocations and patterns of use. In addition, students were required to apply the corpus when preparing their final presentations. During consultation sessions, rather than providing direct answers, guidance was provided for students to return to the corpus to verify and refine their word choices. Post-course interviews revealed that students generally held positive attitudes toward corpus-assisted instruction and perceived it as helpful for transforming passive vocabulary into active use. When asked about future applications, some students suggested that corpus-based instruction could be gradually introduced at lower proficiency levels and recommended the integration of a Chinese–German parallel corpus to reduce learning difficulty for beginners. Overall, by linking a prior teaching experiment with classroom-based practice, this study demonstrates the potential of DDL in Chinese lexical instruction and aims to provide a concrete and pedagogically feasible teaching reference for future Chinese language teachers.