Corpus Methods in Teaching Kazakh as a Foreign Language

Abstract

This thesis project explores the application of corpus-based methods in teaching Kazakh as a foreign language, with a particular focus on the effectiveness of using a spoken language corpus. In recent years, the application of corpora has been widening in the field of language education; and despite the existing research on corpus-informed methods in learning and teaching of English as a foreign language, there remains a gap in such projects in the context of the Kazakh language. Therefore, this work aims to provide strategies for the Kazakh language classes content design enhancements and assess whether corpus-based methods are effective based on the teaching experiment containing a data-driven learning approach (DDL), also known as a discovery method. To achieve these objectives, the Multimedia Corpus of Modern Spoken Kazakh Language (MCSKL) was introduced, and data-driven methods were applied for preparing teaching materials, that are classroom activities and examinations. Afterwards, an experimental study was conducted with the groups of learners in a span of two lessons, each 75 minutes long: a control condition taught through traditional language teaching methods and an experimental condition taught using the corpus-driven techniques. The study utilized post-class assessments, each containing 5 questions of different formats, to analyze the effectiveness of a newly introduced corpus-based instruction through the learners’ performance. Results were finalized in accordance with the objective assessment (selected response) and subjective assessment (point deduction) approaches. The findings revealed that the average final score for a traditional language teaching method was 12.21 points out of 15, which constitutes approximately 81%. In contrast, the average final score for the corpus-informed method was 13.58 out of 15, representing a 10% difference over the previous method and amounting to approximately 91% correct responses. In particular, the objective items (Q1-Q4) show minimal difference between conditions, suggesting comparable learning outcomes on controlled grammatical knowledge. The difference is more pronounced in the productive writing task (Q5), where corpus-informed instruction may have better supported communicative output, although subjective grading also cannot be excluded as a factor. Additionally, in a Likert-scale post-experiment survey, a majority expressed a moderate level of satisfaction with the corpus-driven instruction. However, it is important to note that this study does not represent a mere comparison among the methods and conclude which method is more effective, as the number of participants constitute solely twelve, which cannot function as prototypical for such pedagogical conclusions. Further study should use blind scoring and inter-rater reliability measures for further assessment of the subjective assessment item. Overall, this study provides important insights into Kazakh grammar instruction strategies with an application of spoken corpora.