Learner corpora and the design of data-driven learning activities

Abstract

[EN] This paper seeks to continue the conversation on an underused resource in Data-Driven Learning (DDL), namely learner corpora. It first explores the potential of learner corpora in data-driven learning activity design, focusing on the advantages claimed by a number of scholars throughout the years and mainly associated with the field of Learner Corpus Research (LCR). It then illustrates the status that learner corpus use has in DDL activity design, on the basis of data drawn from the most recent and comprehensive review of DDL research, covering a timespan of 30 years. After describing the main qualitative and quantitative design features of a learner corpus of Italian (i.e. the CELI corpus), the paper shows how learner corpora containing texts produced at different proficiency levels can be used as graded corpora for both target-oriented and error-oriented activities. The sample illustrated activities can thus cater to learners at different proficiency levels, including lower-intermediate and intermediate levels, which are still under-represented in DDL research. Some of the main pedagogical and scientific advantages of using level-specific DDL materials in a paper-based format are also presented and briefly discussed.