A Case Study on Using DeepSeek as an AI-assisted Learning Tool in ESP Reading Education in China

다국어 초록 (Multilingual Abstract)

This study investigated the effectiveness of DeepSeek as an AI-assisted learning tool in English for Specific Purposes (ESP) reading instruction for non-English major university students in China. Motivated by the rapid development of generative artificial intelligence and the growing demand for discipline-oriented English reading competence, this research aims to explore students’ perceptions, learning outcomes, influencing factors, and pedagogical guidelines for integrating AI tools into ESP education. Adopting a mixed-methods sequential explanatory design, the study involved 40 civil engineering undergraduates who participated in a 12-week AI-supported ESP reading program using DeepSeek and the Superstar online platform. Data were collected through needs analysis questionnaires, pre- and post-reading comprehension tests, a post-study perception questionnaire based on the Technology Acceptance Model and Cognitive Load Theory, classroom observation, and semi-structured interviews with both students and teachers. The findings reveal that students hold highly positive perceptions of DeepSeek, particularly regarding its usefulness, ease of use, and ability to reduce cognitive load during academic reading. Quantitative results show significant improvement in students’ ESP reading performance from pre- to post-test scores, especially in vocabulary understanding, text structure recognition, and inferential comprehension. Additional analyses indicate that factors such as digital literacy, learning autonomy, and motivation influence students’ learning gains in AI-assisted environments. Teachers questionnaire and interviews further highlight effective guidelines for integrating AI tools, including prompt literacy training, scaffold reading support, and guidance on critically evaluating AI-generated content. However, challenges such as potential over-reliance on AI and assessment difficulties are also identified.
Overall, this study demonstrates that DeepSeek can serve as an effective and accessible tool to enhance ESP reading competence, promote autonomous learning, and support instructional innovation. The research contributes empirical evidence to the emerging field of AI-assisted ESP education and offers practical guidelines for teachers seeking to integrate AI technologies into future curriculum design.

국문 초록 (Abstract)

본 연구는 중국의 대학생을 대상으로, 특수목적영어(ESP) 수업에서 읽기 학습을 위한 AI 도구 중 하나인 DeepSeek의 효과성을 탐구하였다. 생성형 인공지능의 급속한 발전과 전공 중심 영어 읽기 역량 강화에 대한 요구 증가를 배경으로, 본 연구는 ESP 수업에서 AI 도구를 이용할 때 학습자의 인식과 학습 성과를 분석하고 교수·학습적 지침을 제안하는 것을 목적으로 한다. 연구 방법으로는 혼합 연구를 적용하였으며, DeepSeek와 Superstar 온라인 플랫폼을 활용한 12주간의 AI 지원 ESP 읽기 프로그램에 토목공학 전공 학부생 40명이 참여하였다. 자료는 학생 대상의 요구 분석 설문지, 읽기 이해 사전·사후 시험용 검사지, 기술수용모형과 인지부하이론을 기반으로 한 사후 인식 설문지를 수집하였으며 수업 관찰 진행, 그리고 학생과 교사를 대상으로 한 반구조화 인터뷰을 통해 연구 자료를 수집하여 분석이 이루어졌다. 연구 결과, 학습자들은 DeepSeek에 대해 전반적으로 매우 긍정적인 인식을 보였으며, 특히 유용성, 사용 용이성, 그리고 학술적 읽기 과정에서 인지 부하를 감소시키는 도움 역할에 대해 높게 평가하였다. 양적 분석 결과, 사전 검사와 사후 검사 점수 비교에서 학습자들의 ESP 읽기 성취도가 유의미하게 향상된 것으로 나타났으며, 특히 어휘 이해, 텍스트 구조 인식, 추론적 이해 영역에서 두드러진 향상이 확인되었다. 추가 분석을 통해 디지털 리터러시, 학습 자율성, 학습 동기와 같은 요인이 AI 보조 학습 환경에서 학습자의 성취도 향상에 영향을 미치는 것으로 나타났다. 교사 설문지와 인터뷰 결과는 프롬프트 리터러시 훈련, 단계적 읽기 지원 제공, 그리고 AI 생성 콘텐츠에 대한 비판적 평가를 지도하는 것과 같은 AI 도구 통합을 위한 효과적인 지침을 추가로 제시하였다. 그러나 AI에 대한 과도한 의존 가능성과 평가의 어려움과 같은 한계점 또한 확인되었다. 종합적으로, 본 연구는 DeepSeek가 ESP 읽기 역량을 향상시키고, 학습자의 자율적 학습을 촉진하며, 교수·학습 혁신을 지원하는 효과적이고 접근 가능한 도구로 기능할 수 있음을 입증하였다. 본 연구는 AI를 보조 도구로 이용한 ESP 교육이라는 신흥 연구 분야에 실증적 근거를 제공하며, 향후 유사한 교육과정 설계에서 AI 기술을 통합하고자 하는 교사들에게 실천적 지침을 제시한다.

목차 (Table of Contents)

• I. Introduction 1
• 1.1 Background of the Study 1
• 1.2 Purpose of the Study 4
• 1.3 Statements of Research Questions 6
• II. Literature Review 11
• 2.1 English for Specific Purpose (ESP). 11
• 2.1.1 The Concept of ESP 11
• 2.1.2 English Academic Reading in ESP Education 13
• 2.1.3 Roles and Guidelines in ESP Education 15
• 2.2 Technology Enhanced Language Learning in ESP Education. 18
• 2.2.1 Computer-assisted Language Learning (CALL) 18
• 2.2.2 History and Development of CALL 19
• 2.3 AI-based ESP Education 22
• 2.3.1 Digital Literacy and Technical Skills. 22
• 2.3.2 Motivation, Attitude, and Learning Autonomy. 24
• 2.4 DeepSeek and ESP Education. 26
• 2.4.1 Previous Research on AI-assisted ESP Education 27
• 2.4.2 Opportunities of AI Tools such as DeepSeek in ESP Education 29
• III. Research Methodology 33
• 3.1 Research Design. 33
• 3.2 Research Participants 36
• 3.2.1 Student Participants. 36
• 3.2.2 Teacher Participants 39
• 3.2.3 Sampling Method and Sample Characteristics. 40
• 3.3 Research Instruments 42
• 3.3.1 Pre-study Needs Analysis Questionnaire. 42
• 3.3.2 Reading Comprehension Tests (Pre-test and Post-test) 45
• 3.3.3 Post-study Perception Questionnaire. 50
• 3.3.4 Teachers Questionnaire. 53
• 3.3.5 Semi-structured Interviews. 55
• 3.4 Research Procedures 60
• 3.4.1 Pre-intervention Phase 60
• 3.4.2 Intervention Phase 61
• 3.4.3 Post-intervention Phase 62
• 3.5 Data Collection and Analysis 64
• 3.5.1 Quantitative Data Analysis. 65
• 3.5.2 Qualitative Data Analysis 71
• 3.5.3 Integration of Quantitative and Qualitative Findings. 75
• 3.6 Ethical Consideration 76
• 3.7 Validity and Reliability 77
• 3.7.1 Quantitative Validity and Reliability. 78
• 3.7.2 Qualitative Trustworthiness 80
• 3.7.3 Mixed-methods Integration Quality 82
• IV. Results and Discussion. 83
• 4.1 Research Question 1: Students’ Perceptions of DeepSeek-assisted ESP Reading 83
• 4.1.1 Descriptive Statistics: Technology Acceptance Model Constructs. 84
• 4.1.2 Descriptive Statistics：Cognitive Load Theory Constructs. 88
• 4.1.3 Relationships among Constructs: Correlation and Regression Analysis 92
• 4.1.4 Qualitative Findings: Thematic Analysis of Students Interviews. 98
• 4.1.5 Integrated Interpretation: Synthesizing Quantitative and Qualitative Findings. 110
• 4.2 Research Question 2: Additional Factors Analysis 113
• 4.3 Research Question 3: Learning Outcomes in DeepSeek- assisted ESP Reading. 119
• 4.3.1 Overall Reading Comprehension Outcomes 120
• 4.3.2 Cognitive Level Analysis: Differential Improvement across Reading Skills 123
• 4.3.3 Qualitative Evidence: Mechanisms Underlying Improvement 128
• 4.3.4 Integrated Interpretation: Convergence of Evidence 133
• 4.4 Research Question 4: Teaching Guidelines in AI-assisted ESP Education 134
• 4.4.1 Teachers Questionnaire Results. 135
• 4.4.2 Qualitative Findings: Effective Teaching Guidelines. 139
• 4.4.3 Challenges Encountered and Solutions Developed 146
• 4.4.4 Teacher Professional Learning and Development 149
• 4.4.5 Pedagogical Implications and Recommendations. 151
• 4.5 Chapter Summary 154
• V. Conclusion and Implications. 158
• 5.1 Major Findings. 158
• 5.2 Pedagogical Implications for ESP Education 162
• 5.2.1 Reconceptualizing the Teacher’s Role in AI-augmented ESP Classrooms. 162
• 5.2.2 Curriculum Design Principles for AI-integrated ESP Courses 164
• 5.2.3 Assessment Redesign for AI-augmented ESP Competence 166
• 5.2.4 Equity and Access in AI-enhanced ESP Education. 165
• 5.2.5 Institutional Support for Sustainable AI Integration 167
• 5.3 Theoretical Contributions 168
• 5.3.1 Integrating TAM and CLT in Educational AI Research. 168
• 5.3.2 Elaborating ESP Theory for AI-augmented Learning Contexts 170
• 5.3.3 Reconceptualizing Needs Analysis in AI-enhanced Theory. 171
• 5.3.4 Contributing to Frameworks of Critical AI Literacy 172
• 5.4 Limitations and Future Research Directions. 173
• 5.4.1 Methodological Limitations. 174
• 5.4.2 Contextual Limitations 175
• 5.4.3 Novelty Effect or Hawthorne Effect Consideration. 176
• 5.4.4 Unanswered Questions and Future Directions 177
• 5.5 Reflections 179
• References. 183
• Appendices 206
• Abstract 225
• 국문 초록. 227