Corpus investigation into lexical features of AI-generated texts - vocabulary profiling for A2-level ESL learners

Aug 31, 2026

Upload by Jacob

A practice-oriented educational research project investigates how common generative AI tools can generate reading materials for A2-level ESL learners. The study conducts a corpus investigation into lexical features of AI-generated texts using vocabulary profiling and measurement tools, including EVP-, GSL-, AWL-word lists and ignore lists. Texts are compared with handbook materials and Key A2 reference texts, with additional comparisons of different AI tools. Results analyze quantitative differences and discuss pedagogical implications, limitations, and directions for further research.

LanguageEnglish

CategoryPaper Templates

FormatDOCX

Pages130

Reading Time46 min

AccessFree to download

Abstract – English

Abstract – Dutch

Introduction

Literature Review

Evolutions of gen-AI

Defining A2-level

A2-level in Flemish education

Reading comprehension thresholds

Research gaps and research questions

Methodology

Vocabulary profiling and measurement tools

1. EVP-lists

2. GSL-lists

3. AWL-list

4. Ignore lists

Corpus design and selection

Key A2

Handbooks

AI-based text generation: tools and procedures

Tool selection

Prompt engineering

Challenges with AI-generated texts

Results

Discussion

Conclusion

References

Using common gen-AI tools to generate reading materials for A2-level ESL learners. A CORPUS INVESTIGATION INTO LEXICAL FEATURES OF AI-GENERATED TEXTS Word count: 9365 First name: Annelies Surname: Pandelaers Student number: 02408201 Supervisor(s): Prof. Dr. Maribel Montero Perez Practice Oriented Educational Research Project (9SP) submitted to Ghent University in partial fulfillment of the requirements for the degree of Master of Teaching in Languages. Academic year: 2024 – 2025, Master of Teaching Table of Contents Inhoudsopgave TOC \o "1-3" \h \z \u HYPERLINK \l "_Toc205813299" Table of Contents PAGEREF _Toc205813299 \h 2 HYPERLINK \l "_Toc205813300" List abbreviations PAGEREF _Toc205813300 \h 6 HYPERLINK \l "_Toc205813301" List of tables and figures PAGEREF _Toc205813301 \h 7 HYPERLINK \l "_Toc205813302" Abstract – English PAGEREF _Toc205813302 \h 8 HYPERLINK \l "_Toc205813303" Abstract – Dutch PAGEREF _Toc205813303 \h 9 HYPERLINK \l "_Toc205813304" Introduction PAGEREF _Toc205813304 \h 10 HYPERLINK \l "_Toc205813305" Literature Review PAGEREF _Toc205813305 \h 11 HYPERLINK \l "_Toc205813306" Evolutions of gen-AI PAGEREF _Toc205813306 \h 11 HYPERLINK \l "_Toc205813307" Defining A2-level PAGEREF _Toc205813307 \h 12 HYPERLINK \l "_Toc205813308" A2-level in Flemish education PAGEREF _Toc205813308 \h 14 HYPERLINK \l "_Toc205813309" Reading comprehension thresholds PAGEREF _Toc205813309 \h 15 HYPERLINK \l "_Toc205813310" Research gaps and research questions PAGEREF _Toc205813310 \h 16 HYPERLINK \l "_Toc205813311" Methodology PAGEREF _Toc205813311 \h 17 HYPERLINK \l "_Toc205813312" Vocabulary profiling and measurement tools PAGEREF _Toc205813312 \h 17 HYPERLINK \l "_Toc205813313" 1. EVP-lists PAGEREF _Toc205813313 \h 17 HYPERLINK \l "_Toc205813314" 2. GSL-lists PAGEREF _Toc205813314 \h 18 HYPERLINK \l "_Toc205813315" 3. AWL-list PAGEREF _Toc205813315 \h 18 HYPERLINK \l "_Toc205813316" 4. Ignore lists PAGEREF _Toc205813316 \h 18 HYPERLINK \l "_Toc205813317" Corpus design and selection PAGEREF _Toc205813317 \h 19 HYPERLINK \l "_Toc205813318" Key A2: PAGEREF _Toc205813318 \h 19 HYPERLINK \l "_Toc205813319" Handbooks: PAGEREF _Toc205813319 \h 20 HYPERLINK \l "_Toc205813320" AI-based text generation: tools and procedures PAGEREF _Toc205813320 \h 21 HYPERLINK \l "_Toc205813321" Tool selection PAGEREF _Toc205813321 \h 21 HYPERLINK \l "_Toc205813322" Prompt engineering PAGEREF _Toc205813322 \h 21 HYPERLINK \l "_Toc205813323" Challenges with AI-generated texts PAGEREF _Toc205813323 \h 24 HYPERLINK \l "_Toc205813324" Conclusion PAGEREF _Toc205813324 \h 25 HYPERLINK \l "_Toc205813325" Results PAGEREF _Toc205813325 \h 26 HYPERLINK \l "_Toc205813326" Explanation of data points PAGEREF _Toc205813326 \h 26 HYPERLINK \l "_Toc205813327" Comparison of handbook and Key A2 texts versus AI-generated texts PAGEREF _Toc205813327 \h 27 HYPERLINK \l "_Toc205813328" Comparison handbooks only vs AI PAGEREF _Toc205813328 \h 28 HYPERLINK \l "_Toc205813329" Comparison ChatGPT vs Gemini PAGEREF _Toc205813329 \h 28 HYPERLINK \l "_Toc205813330" Summary of key quantitative findings PAGEREF _Toc205813330 \h 29 HYPERLINK \l "_Toc205813331" Conclusion PAGEREF _Toc205813331 \h 30 HYPERLINK \l "_Toc205813332" Discussion PAGEREF _Toc205813332 \h 31 HYPERLINK \l "_Toc205813333" Connection to prior research and theoretical models PAGEREF _Toc205813333 \h 31 HYPERLINK \l "_Toc205813334" Interpretation of findings PAGEREF _Toc205813334 \h 31 HYPERLINK \l "_Toc205813335" AI-tool specific analysis PAGEREF _Toc205813335 \h 32 HYPERLINK \l "_Toc205813336" Practical implications & limitations PAGEREF _Toc205813336 \h 33 HYPERLINK \l "_Toc205813337" Summary of key insights

FAQs

研究的核心问题是什么？

研究聚焦于生成式AI生成的文本在词汇层面的特征，并以A2级ESL学习者阅读材料为对象开展语料库调查与比较分析。

研究如何测量文本的词汇特征？

通过词汇画像与测量工具进行分析，使用EVP、GSL、AWL词表以及忽略列表等方法对词汇分布进行统计与对比。

研究的比较对象有哪些？

AI生成文本会与教材手册文本与Key A2参考文本进行比较，同时还比较不同AI工具生成的文本差异。

Aug 31, 2026

Upload by Jacob