The EuroCALL Review - No. 2 - Comparing ChatGPT and Authentic Discourse as L2 Academic Speaking Partners

Oct 15, 2026

Uploaded by Ophelia

This study examines the extent to which spoken academic English produced by the ChatGPT-4o model aligns with authentic discourse and is applicable for L2 speaking materials. Ten dialogues generated by ChatGPT were compared with an authentic transcript of a doctoral dissertation defence from MICASE using corpus linguistic and content analysis. Results show higher lexical diversity, advanced vocabulary, and more complex syntax, but lower readability. ChatGPT simulated turn-taking and question–answer patterns yet lacked spontaneous-interaction features such as clarification requests and overlapping speech.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages13

Reading Time33 min

AccessFree to download

Introduction

Abstract

Keywords

The EuroCALL Review Volume 32, No. 2, 2025, 75-87 ISSN: 1695-2618 [https://doi.org/10.4995/eurocall.2025.23842](https://doi.org/10.4995/eurocall.2025.23842) RESEARCH PAPER Comparing ChatGPT and Authentic Discourse as L2 Academic Speaking Partners: A Corpus-based Analysis and Exploration of Pedagogical Applicability Koki Sekitani *; Masaaki Ogura **; Takeshi Sato ** * *Toyo Eiwa University, Japan; ** Keio University, Japan; *** Kobe University, Japan *[sekitani.koki@toyoeiwa.ac.jp](sekitani.koki@toyoeiwa.ac.jp |)[ |](sekitani.koki@toyoeiwa.ac.jp |) ** [m-ogura@keio.jp](m-ogura@keio.jp |)[ |](m-ogura@keio.jp |) ***[satoken@people.kobe-u.ac.jp](satoken@people.kobe-u.ac.jp) How to cite this article: Sekitani, K., Ogura, M., & Sato, T. (2025) . Comparing ChatGPT and Authentic Discourse as L2 Academic Speaking Partners: A Corpus-based Analysis and Exploration of Pedagogical Applicability. The EuroCALL Review, 32 (2), 75-87. [https://doi.org/10.4995/eurocall.2025.23842](https://doi.org/10.4995/eurocall.2025.23842) Abstract This study examines the extent to which spoken academic English produced by the ChatGPT-4o model (ChatGPT) aligns with authentic discourse and is applicable for use in L2 speaking materials. As generative AI tools gain popularity among L2 learners for speaking practice, it is essential to assess the linguistic and discourse-level validity of such output. Ten dialogues generated by ChatGPT were compared with an authentic transcript of a doctoral dissertation defence from the Michigan Corpus of Academic Spoken English (MICASE) using a corpus linguistic and content analysis approach. Quantitative analyses revealed that ChatGPT produced language with higher lexical diversity, greater use of advanced vocabulary, and more complex syntax than the MICASE sample, but with lower readability. Content analysis showed that while ChatGPT simulated turn-taking and appropriate question–answer sequences, it lacked features of spontaneous interaction such as clarification requests, topic shifts, and overlapping speech. Speaker roles in the generated texts were consistent but followed more scripted and idealized patterns. These findings suggest that ChatGPT can serve as a useful supplementary tool for academic speaking practice, especially in providing structured, high-level input. However, its limitations in simulating authentic interaction highlight the need for guided integration alongside real human dialogue. The study is expected to contribute to the growing body of research on AI-assisted language learning. Keywords ChatGPT; academic spoken English; corpus linguistic analysis; L2 speaking practice; syntactic complexity; readability 1. Introduction This study examines the authenticity of academic spoken English discourse generated by ChatGPT and its applicability to L2 learning activities by comparing it with discourse from an academic spoken English corpus. ChatGPT has been rapidly and widely adopted by L2 learners to enhance their proficiency and fluency. Several studies (e.g., Yang & Li, 2024) have investigated the impact of ChatGPT on L2 learning with respect to writing. More specifically, Su et al. (2023) showed that ChatGPT provides valuable feedback at various stages of the L2 writing process, such as brainstorming, revision, and proofreading. Mizumoto et al. (2024) also reported that writing assessments with ChatGPT demonstrate comparable reliability to those of humans. Beyond writing, emerging research suggests that ChatGPT can serve as an effective partner for L2 oral interaction (e.g., Hayashi & Sato, 2024) . It allows learners to speak with reduced pressure and anxiety (Javier & Moorhouse, 2023; Hayashi & Sato, 2024) , alleviating fears of making errors or receiving negative feedback that might hinder performance (Gregersen, 2023) . By providing elaborate and contextually relevant responses almost instantly, ChatGPT simulates dynamic and interactive conversational experiences

FAQs

What is the main purpose of the study?

The study assesses how closely ChatGPT-generated spoken academic English matches authentic academic discourse and whether it can be used in L2 speaking materials.

How were the ChatGPT dialogues evaluated against authentic discourse?

Ten ChatGPT-generated dialogues were compared with an authentic transcript from MICASE using corpus linguistic analysis and content analysis.

What linguistic differences were found between ChatGPT output and MICASE?

ChatGPT showed higher lexical diversity, greater use of advanced vocabulary, and more complex syntax, but it had lower readability.

What limitations did the study identify in ChatGPT as an interaction partner?

Although it simulated turn-taking and appropriate question–answer sequences, it lacked spontaneous interaction features such as clarification requests, topic shifts, and overlapping speech.

Oct 15, 2026

Uploaded by Ophelia