Second language (L2) learner corpora are collections of language samples that demonstrate learners’ abilities to perform some learning tasks, e.g. an ability to write essays, answer to reading comprehension questions, or talk on a given topic. Such corpora are necessary for both empirical-based research within Second Language Acquisition (SLA), and for development of methods for automatic processing of such data. L2 corpora are notoriously difficult to collect, and their value depends to a greater degree on the representativeness and balance of the sampled data, type of associated metadata and reliability of manual annotations. In this chapter we thoroughly describe the SweLL-gold corpus of L2 Swedish, its annotation, statistics and metadata, and showcase main types of its use, such as (1) in research on SLA through detailed instructions on how to perform corpus searches given SweLL-specific annotation, combined with guidelines for SVALA usage, a tool for correction annotation; and (2) in NLP research on problems such as grammatical error correction through guidelines on how to use the different available file formats that the SweLL-gold corpus is released in. Both cases are further supported by case studies and, where available, relevant scripts ready for reuse by researchers.

This article on learner corpus design presents a novel methodological approach, founded on a systematic and clear separation between normalization (‘reconstruction’, ‘standardization’) and correction annotation (‘error annotation’), where the correction annotation is based on a predefined normalized version of each learner text, and where the normalized texts are presented and accessible as a separate corpus, parallel to the corpus of the original learner texts. It is argued that the nature of the normalization process, as distinct from the correction annotation, requires a holistic and interpretation-based approach, which increases the validity of the correction annotation as an analysis of differences between 1) the writer’s choice of expression, and 2) a normadhering way of formulating the message intended by the writer. The case in point is the recently released Swedish learner language corpus (SweLL, Volodina et al. 2019).

Språkbanken lanserade 2021 en ny annoterad digital inlärarkorpus med texter skrivna av inlärare av svenska, SweLL (Swedish Learner Language corpus). Syftet med artikeln är att diskutera och ge exempel på hur digitala inlärarkorpusar kan bidra till språkforskningen om inlärares språkanvändning, vilket är en fråga som fått allt mer uppmärksamhet (Le Bruyn & Paquot, 2021; Myles, 2015). Genom att utgå ifrån diskussionen om relationen mellan forskningsfälten LCR (Learner Corpus Research) och SLA (Second Language Acquisition) kommer vi i artikeln att presentera den nya inlärarkorpusen och diskutera några viktiga teoretiska och metodologiska aspekter i samband med dess utveckling (Volodina m.fl., 2019).Vi ger även exempel på några tillämpningar av korpusen genom en explorativ studie av förflyttningsverbkonstruktioner i inlärartexterna. Resultaten på gruppnivå stärker tidigare forskning som visar en överrepresentation av verbkonstruktioner med gå hos tidiga inlärare.

The article presents a new language learner corpus for Swedish, SweLL, and the methodology from collection and pesudonymisation to protect personal information of learners to annotation adapted to second language learning. The main aim is to deliver a well-annotated corpus of essays written by second language learners of Swedish and make it available for research through a browsable environment. To that end, a new annotation tool and a new project management tool have been implemented, both with the main purpose to ensure reliability and quality of the final corpus. In the article we discuss reasoning behind metadata selection, principles of gold corpus compilation and argue for separation of normalization from correction annotation.

This paper reports on the status of learner corpus anonymization for the ongoing research infrastructure project SweLL. The main project aim is to deliver and make available for research a well-annotated corpus of essays written by second language (L2) learners of Swedish. As the practice shows, annotation of learner texts is a sensitive process demanding a lot of compromises between ethical and legal demands on the one hand, and research and technical demands, on the other. Below, is a concise description of the current status of pseudonymization of language learner data to ensure anonymity of the learners, with numerous examples of the above-mentioned compromises.