An Editable Learner Model for Text Recommendation for Language Learning

Autor/in	Lee, John S. Y.
Titel	An Editable Learner Model for Text Recommendation for Language Learning
Quelle	In: ReCALL, 34 (2022) 1, S.51-65 (15 Seiten)Infoseite zur Zeitschrift PDF als Volltext Verfügbarkeit
Zusatzinformation	ORCID (Lee, John S. Y.)
Sprache	englisch
Dokumenttyp	gedruckt; online; Zeitschriftenaufsatz
ISSN	0958-3440
DOI	10.1017/S0958344021000197
Schlagwörter	Second Language Learning; Reading Materials; Computer Assisted Instruction; Second Language Instruction; Individualized Instruction; Chinese; Difficulty Level + Suchen Sie Ihr Suchwort? Zweitsprachenerwerb; Computer based training; Computerunterstützter Unterricht; Fremdsprachenunterricht; Individualisierender Unterricht; China; Chinesen; Schwierigkeitsgrad
Abstract	Extracurricular reading is important for learning foreign languages. Text recommendation systems typically classify users and documents into levels, and then match users with documents at the same level. Although this approach can be effective, it has two significant shortcomings. First, the levels assume a standard order of language acquisition and cannot be personalized to the users' learning patterns. Second, recommendation decisions are not transparent because the leveling algorithms can be difficult for users to interpret. We propose a novel method for text recommendation that addresses these two issues. To enhance personalization, an open, editable learner model estimates user knowledge of each word in the foreign language. The documents are ranked by new-word density (NWD) -- that is, the percentage of words that are new to the user in the document. The system then recommends documents according to a user-specified target NWD. This design offers complete transparency as users can scrutinize recommendations by reviewing the NWD estimation of the learner model. This article describes an implementation of this method in a mobile app for learners of Chinese as a foreign language. Evaluation results show that users were able to manipulate the learner model and NWD parameters to adjust the difficulty of the recommended documents. In a survey, users reported satisfaction with both the concept and implementation of this text recommendation method. (As Provided).
Anmerkungen	Cambridge University Press. 100 Brook Hill Drive, West Nyack, NY 10994. Tel: 800-872-7423; Tel: 845-353-7500; Fax: 845-353-4141; e-mail: subscriptions_newyork@cambridge.org; Web site: https://www.cambridge.org/core/what-we-publish/journals
Erfasst von	ERIC (Education Resources Information Center), Washington, DC
Update	2024/1/01