Search

The Auxiliary Effect of Large Language Models on the Ability to Expand Topics in IELTS Speaking

Author(s)

Xu Ke

Affiliation(s)

Liaoning University, Shenyang, Liaoning, China

Abstract

An IELTS Speaking band 5.5 represents a typical proficiency bottleneck for intermediate-level test-takers aiming to reach the widely targeted band 6.0 entry requirement of overseas institutions. Test-takers at this band are primarily hindered by insufficient topic expansion ability, manifested as narrow thinking, repetitive expressions and loose logic. To verify the auxiliary effect of Large Language Models (LLMs) on improving such ability among this group of learners striving for band 6.0, this study is grounded in the Output Hypothesis of second language acquisition proposed by Swain (1985) [2] and Interaction Feedback research reviewed by Nassaji (2020) [3]. Ten English majors with an IELTS Speaking band of 5.5 were selected as research subjects and divided into an experimental group (training with LLM and corpus feedback) and a control group (traditional exam preparation training) for an 8-week comparative experiment. A multi-dimensional feedback system covering breadth of topic coverage, lexical diversity, syntactic richness and logical coherence was constructed, and the effectiveness was verified through mock test scoring and quantitative corpus analysis. The results show that the auxiliary strategy integrating LLMs and corpora can significantly boost the topic expansion capacity of band-5.5 test-takers and precisely address their score bottlenecks toward band 6.0. This study provides operable preparation schemes for intermediate-level IELTS Speaking candidates and enriches empirical research on the application of LLMs in second language oral acquisition.

Keywords

Large Language Models; IELTS Speaking; Topic Expansion; Corpus; Band-5.5 Test-Takers

References

[1] Gui, S. C., & Yang, H. Z. (2003). Chinese Learner English Corpus. Shanghai Foreign Language Education Press. [2] Swain, M. (1985). Communicative competence: Some roles of comprehensible input and comprehensible output in its development. In S. Gass & C. Madden (Eds.), Input in Second Language Acquisition (pp. 235–253). Newbury House. [3] Nassaji, H. (2020). Assessing the effectiveness of interactional feedback for L2 acquisition: Issues and challenges. Language Teaching, 53(2), 159–189. [4] Jones, A., & Smith, B. (2022). Data-driven learning for L2 oral proficiency development. Journal of Corpus Linguistics, 27(3), 45–68. [5] Brown, C. (2023). Topic expansion frameworks in L2 oral English: A corpus-based study. System, 108, 102890. [6] Miller, D., & Lee, S. (2023). GPT-4 assisted academic writing: Logical reconstruction and argument enhancement. Computers & Education, 205, 104567. [7] Garcia, M., & Rodriguez, L. (2024). Scaffolded L2 writing instruction with large language models. Language Learning & Technology, 28(1), 112–130.