Home

Volume 20 Number 3
Abstract

Data-driven Learning of Collocations: Learner Performance, Proficiency, and Perceptions
Nina Vyatkina, University of Kansas

This study explores the effects of Data-Driven Learning (DDL) of German lexico-grammatical constructions (verb-preposition collocations) by North-American college students with intermediate foreign language proficiency. The study compares the effects of computer-based and paper-based DDL activities as evidenced in learners’ immediate and delayed performance gains, and explores changes in learners’ proficiency and DDL perceptions as well as the influence of these factors on performance. The results show that both DDL types were equally effective for all learners, independent of their proficiency and perceptions, although gains measured by a more controlled production test (gap-filling) were superior to and longer lasting than gains measured by a less controlled production test (sentence-writing). Furthermore, immediate performance gains on different tasks were differently affected by learner proficiency and perceptions, while delayed gains showed no such effects. Finally, the study found that overall learner proficiency increased and that DDL was well received by learners and they expressed an intention to use it for independent learning in the future. This study fills gaps existing in DDL research by focusing on a second language other than English, comparing different DDL types, measuring delayed learning gains, and combining different outcomes measures in a multilevel modeling design.

Article PDF