Optimising AI-Based Dialogue Systems for Language Learning in Kazakhstan

Zh.B. Zhauynshiyeva, Balzhan Agabekova, Gulmira Mirkenova, D.T. Ismailova, Zhaxylyk A Oralkanova

Cuadernos de Investigación Filológica, 2026

Full text Open access: Gold Metadata (CSL-JSON)
DOI 10.18172/cif.6714
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

The purpose of this study was to investigate the processes associated with the development and optimisation of chatbots in compliance with language rules. Using linguistic analysis, this study examined the key characteristics of chatbot use, particularly considering the linguistic features of the Kazakh language in the context of teaching English. The stages of chatbot development include analysing user needs, conceptual and structural design, preparation of learning materials, multimedia integration, testing, personalisation, feedback analysis, algorithm improvement, and platform integration. Optimisation steps encompass user data analysis (requests, errors), enhancement of natural language processing (NLP), adaptive content delivery, performance monitoring, and continuous refinement. A survey of 50 students from two Kazakh gymnasium schools revealed that Duolingo was preferred for phonetics (43% and 41%), grammar (53% and 59%), and vocabulary (48% and 39%). Mondly and HelloTalk were favoured for syntax (36-39%), while Mondly, HelloTalk, and Tandem were most effective for constructing coherent speech (35-46%). These results underscore the varying strengths of chatbots across linguistic domains. Optimising Kazakh-language chatbots such as Soyle Bot, Bilim Class, and Qazaq Grammar Chat requires consideration of the agglutinative structure, morphological richness, and dual alphabet systems (Latin and Cyrillic). Effective strategies include gamification, multimedia integration, content localisation, alignment with educational standards, and personalisation.