Development and Prospects of National Intelligent System for Testing General Language Competencies Deployed Through Neural Network Solutions

Abstract

The article presents the results of approbation of the intellectual system for testing general language competences (ISTOK) developed by testologists, linguists, specialists in methodology of teaching foreign languages and in artificial intelligence. This system includes a range of tests assessing language ability at levels from A2 to C1 of the Common European Framework of Reference (CEFR), as well as an adaptive placement test. All test materials are calibrated according to the CEFR. ISTOK is an adaptive testing system deployed through neural network solutions and providing assessment of receptive and productive language skills (reading, listening, speaking and writing) by using artificial intelligence and/or neurolinguistic models. The process of ISTOK development implied, apart from writing test items, putting together databases of writing and speaking assignments marked by professional assessors and assessment criteria for productive skills, as well as algorithms to identify various types of mistakes with the help of artificial intelligence. The results of various testing cohorts with the total number of test takers exceeding 5,000 demonstrated high reliability and objectified test validity. The new approach to language skills testing can be used for various purposes in higher education institutions, as well and to identify and/or confirm language proficiency of personnel in different organisations and businesses, while the principles of training and practical use of neurolinguistic models will find wide application in various fields of applied research.

Keywords

• language testing
• language assessment
• language skills testing
• adaptive testing
• machine learning
• neuro-linguistic model
• linguistic corpus
• databases
• neural network solutions
• intellectual system of testing common language competencies