Assessing automatic text classification for interactive language learning

Author

Branco, Antonio ; Rodrigues, Jose ; Costa, Francois ; Silva, Jaime ; Vaz, Richard

Author_Institution

Dept. of Inf., Univ. of Lisbon, Lisbon, Portugal

fYear

2014

fDate

10-12 Nov. 2014

Firstpage

70

Lastpage

78

Abstract

In this paper we discuss the design options for a language processing tool that supports humans in their task of classifying text excerpts according to CEFR levels of language proficiency. We describe the tool that we developed on the basis of these design options and provide an assessment of its functioning. This tool is suitable to be used by students taking courses of Portuguese as a second language, as well as by expert instructors selecting items of exam that are aimed at assessing and certifying the language level of these students. It is an instrument that aims at supporting humans in their language level classification judgments by providing conditions to more consistent and objectively sustained judgments across different occasions and input texts and across different human classifiers. Its design principles and underlying language technology can be applied to develop similar tools for any other language.

Keywords

computer aided instruction; distance learning; educational courses; natural language processing; pattern classification; text analysis; MOOC; Portuguese language course; automatic text classification; e-learning; interactive language learning; language processing tool; massive open online course; Accuracy; Certification; Indexes; Instruments; Measurement; Natural language processing; Reliability; CEFR; Portuguese language; language learning assessment; readability assessment;

fLanguage

English

Publisher

ieee

Conference_Titel

Information Society (i-Society), 2014 International Conference on

Conference_Location

London

Type

conf

DOI

10.1109/i-Society.2014.7009014

Filename

7009014