Impact of ChatGPT on EFL Writing Anxiety & Idea Generation
Academic study exploring how ChatGPT-assisted brainstorming affects idea generation and reduces writing anxiety among Master's students in English.
PEOPLE'S DEMOCRATIC REPUBLIC OF ALGERIA
MINISTRY OF HIGHER EDUCATION AND SCIENTIFIC RESEARCH
MOHAMED KHEIDER UNIVERSITY – BISKRA
Faculty of Letters and Foreign Languages
Dept. of Language & English Literature
Academic Year: 2025/2026
Master's Dissertation Defense
Investigating the Impact of ChatGPT-Assisted Brainstorming on EFL Learners' Idea Generation and Writing Anxiety
The Case of Master's Students at Biskra University
Ali MJENNAH
Dr. Messaouda BENDAHMANE
Board of Examiners
Dr. Fatima MESSAOUDI (Chairperson)
Dr. Messaouda BENDAHMANE (Supervisor)
Mr. Khaled AMRAOUI (Examiner)
Mrs. Thaldja AICHAOUI (Examiner)
Outline
Presentation Overview
Outline
Introduction & Background
Statement of the Problem
Aim & Objectives
Research Questions
Research Hypotheses
Methodology
Results & Findings
Discussion & Synthesis
Recommendations
Conclusion
Introduction
Section 01
Introduction
Background of the Study
Writing is a critical and complex skill for EFL learners, involving linguistic, cognitive, and psychological challenges simultaneously.
The pre-writing stage is particularly demanding — students struggle with both idea generation and foreign language writing anxiety, which intensify each other.
In the Algerian EFL context, these obstacles are especially significant for Master's students who must produce advanced academic texts and write their own dissertations.
The rise of Generative AI tools — particularly ChatGPT — presents a new opportunity to address these challenges in language learning contexts.
"Writing is an emotional as well as a cognitive activity — we think and feel while we are writing." (Cheng, 2002)
Problem
Section 02
Statement of the Problem
EFL learners face significant difficulties during the pre-writing stage — particularly in generating and organizing ideas — which triggers FL Writing Anxiety.
In the Algerian context, this is especially critical for Master's students at Biskra University who must produce advanced academic writing for their degrees.
Traditional brainstorming techniques may no longer be sufficient for these advanced learners — a new, evidence-based strategy is needed.
While ChatGPT has shown potential globally as a scaffolding tool, its specific dual impact on idea generation and writing anxiety in the Algerian EFL context remains unexplored.
This study fills a critical gap by empirically investigating ChatGPT's dual effect on EFL learners' cognitive and affective writing challenges in an under-researched context.
Aim & Objectives
Section 03
Aim & Objectives
Main Research Aim
This study aims to investigate the impact of ChatGPT-assisted brainstorming on idea generation and writing anxiety among EFL Master's students at Biskra University, and to evaluate ChatGPT as an evidence-based alternative to traditional brainstorming techniques.
Specific Objectives
Examine the impact of ChatGPT-assisted brainstorming on EFL students' idea generation.
Analyze the effect of ChatGPT-assisted brainstorming on EFL students' writing anxiety levels.
Explore EFL Master's students' perceptions of ChatGPT as a brainstorming assistant.
RQs & RHs
Section 04 & 05
Research Questions & Hypotheses
Research Questions
RQ1
To what extent does ChatGPT-assisted brainstorming enhance EFL Master's students' idea generation?
RQ2
How does ChatGPT-assisted brainstorming influence EFL students' writing anxiety?
RQ3
What are the perceptions of EFL Master's students regarding ChatGPT as a brainstorming tool?
Research Hypotheses
RH1
ChatGPT-assisted brainstorming has no significant effect on idea generation.
Significantly enhances it.
RH2
ChatGPT-assisted brainstorming has no significant effect on writing anxiety.
Significantly reduces it.
RH3
EFL Master's students display positive perceptions toward ChatGPT as a brainstorming tool.
Methodology
Section 06
Methodology
Research Paradigm
Pragmatic paradigm — justifying mixed methods to solve real-world problems.
Research Approach
Explanatory Sequential Mixed-Methods Design (QUAN → qual).
Sample
<div style='margin-bottom:12px;'>N = 28 EFL Master One students (convenience sampling) + 6 for focus group (purposive sampling).</div><div style='font-style:italic; color:#b5832a; font-weight: 600;'>Biskra University, 2025/2026</div>
Instruments
<div style='display:flex; flex-direction:column; gap:10px;'><div><b>(1)</b> Adapted SLWAI — 22-item Likert scale (writing anxiety).</div><div><b>(2)</b> Idea Generation Tasks A & B — scored via analytic rubric (quantity, diversity, relevance).</div><div><b>(3)</b> Focus Group Interview.</div></div>
Data Analysis
<div style='display:flex; flex-direction:column; gap:14px;'><div><b style='color:#1a5c38;'>Quantitative:</b> IBM SPSS 27 — descriptive stats, paired-samples t-test, Cohen's d.</div><div><b style='color:#1a5c38;'>Qualitative:</b> Thematic analysis via MAXQDA.</div></div>
Intervention (Treatment)
<div style='display:flex; flex-direction:column; gap:10px;'><div>• 5 sessions over 2 weeks.</div><div>• SCAFFOLD model + structured prompting strategies.</div><div style='color:#b5832a; font-weight:700; margin-top: 5px; font-style: italic;'>Pre-test → Intervention → Post-test → FGD</div></div>
Results: Idea Generation
Section 07 — Quantitative Results
Results: Idea Generation Task
Section A — Descriptive Statistics
Metric
Pre-Test M
Post-Test M
Quantity
1.71
2.00 (↑)
Relevance
2.54
2.18 (↓)
Diversity
2.00
2.32 (↑)
Total Score
6.25
6.50 (↑)
Section B — Inferential Statistics
Paired t-Test:
t(27) = 0.72, p = .478 (> .05)
→ NOT statistically significant
Effect Size:
Cohen's d = 0.14
→ Very Small Effect
ChatGPT-assisted brainstorming did not significantly improve idea generation statistically, though descriptive improvements in quantity and diversity were observed.
Results: Writing Anxiety
Section 07 — Quantitative Results
Results: Writing Anxiety (SLWAI)
Section A — Descriptive Statistics
Subscale
Pre-Test M
Post-Test M
Change
Somatic Anxiety
27.71
23.71
↓ 4.00
Cognitive Anxiety
← largest per-item reduction
21.68
18.18
↓ 3.50
Avoidance Behavior
12.46
12.00
↓ 0.46
Total SLWAI
61.86
53.89
↓ 7.97
Low anxiety: 21.4% → 46.4% ✓ | High anxiety: 35.7% → 25% ✓
Section B — Inferential Statistics
Paired t-Test: t(27) = -3.491, p = .002 (< .05) → STATISTICALLY SIGNIFICANT ✓
Effect Size: Cohen's d = -0.66 → Medium Effect
ChatGPT-assisted brainstorming significantly reduced EFL learners' writing anxiety, with cognitive anxiety showing the largest reduction. RH2(0) rejected.
Qualitative Results
Section 07 — Qualitative Results
Results: Focus Group Interview
Thematic Analysis — 4 Themes Identified
ChatGPT as a Facilitator of Idea Generation
Participants reported improvements in idea quantity, quality, diversity, and logical organization. ChatGPT helped them "think out of the box."
Reduction of Writing Anxiety & Increased Confidence
Students described feeling more confident, motivated, and less stressed. ChatGPT created a psychologically safe writing environment.
Ethical Use Through Structured Scaffolding & Prompting
The SCAFFOLD model and prompting strategies were praised as essential for ethical, principled use. Students shifted from ghostwriter to brainstorming-assistant mentality.
Challenges & Limitations of ChatGPT Use
Concerns raised about over-reliance, plagiarism risks, accessibility issues, and potential loss of creativity and personal writing style.
Recommendations
Section 08
Pedagogical Implications & Recommendations
For Policymakers
Clearly define the ethical use of AI in education; ensure accessibility of AI tools in universities; encourage principled AI integration in language learning.
For Curriculum Designers
Develop and integrate AI-assisted writing initiatives into EFL curricula; design programs that support the SCAFFOLD model approach.
For Teachers
Shift from viewing ChatGPT as a competitor to embracing it as a collaborator; use it to foster both cognitive and emotional engagement in writing classrooms.
For Students
Avoid over-reliance; use ChatGPT as a thinking partner, not a shortcut; maintain critical thinking and personal writing voice.
The key takeaway: Ethical, structured AI integration — not prohibition — is the path forward for EFL writing instruction.
Conclusion
Section 09
Conclusion
Summary of Findings
ChatGPT-assisted brainstorming produced a positive but non-significant improvement in idea generation (p = .478). Qualitative data, however, revealed meaningful perceived improvements in quantity, diversity, and organization.
ChatGPT significantly reduced EFL learners' FL writing anxiety (p = .002, d = -0.66, medium effect). Cognitive anxiety showed the largest reduction.
Students displayed overwhelmingly positive perceptions, praising the SCAFFOLD model and prompting strategies as essential for ethical and effective ChatGPT use.
Dissertation Structure
Chapter One reviews the theoretical framework: EFL writing, brainstorming, writing anxiety, and AI in language learning. Chapter Two presents the methodology, data analysis, and a synthesized discussion of quantitative and qualitative findings.
When implemented ethically and structurally, ChatGPT can serve as a powerful scaffold that simultaneously addresses the cognitive and affective challenges of EFL writing — offering a pedagogically sound path for AI integration in language instruction.
Mohamed Kheider University of Biskra
2025 / 2026
Thank You
The board of examiners' questions and feedback are greatly welcomed.
Submitted by:
Ali MJENNAH
Supervised by:
Dr. Messaouda BENDAHMANE
Keywords: AI · Brainstorming · ChatGPT · Foreign Language Writing Anxiety · Idea Generation
- chatgpt-in-education
- efl-writing
- academic-research
- writing-anxiety
- brainstorming-tools
- ai-literacy
- mixed-methods-study