Abstrakt
Research on learner corpora (LC) is today a full-fledged discipline and has established itself as an indispensable research methodology applied to the study of learning varieties. This contribution discusses the state of the art of Learner Corpus Research, with particular reference to the analysis trends and discussion topics that have characterized it during its evolution - from a more theoretical-planning first phase to more contemporary approaches operational and user-oriented – and to the main characteristics of the Italian LS/L2 LCs.
Among the elements that make LCs useful and exploitable on several levels by different users there is certainly the labeling of errors, an age-old and complex issue that has animated the debate around these resources since the dawn of their creation. In this regard, the VALICO-UD corpus (subcorpus of the VALICO LC) is presented, a resource created by applying the Universal Dependencies framework and an error labeling tagset to a portion of VALICO; its properties, criticalities, choices and methodology followed are illustrated.
Finally, we move on to consider how the LCs from tools and objects of research can become tools and objects of teaching LS/L2 Italian, referring to data-driven learning experiences and corpus-based research-action projects.