ChatGPT as Third Space: Student Agency and Teacher Insight Through AI-Mediated Literacy Practices
ChatGPT as Third Space: Student Agency and Teacher Insight Through AI-Mediated Literacy Practices
By Maria Lisak, EdD (How to cite this)
Abstract
This paper explores university students’ engagement with ChatGPT as a third space for literacy, identity negotiation, and agency. Drawing on sociocultural theory (Lewis, Enciso, & Moje, 2020) and third space theory (Gutiérrez, 2008), the study examines student reflections on their use of ChatGPT for academic and personal purposes in an English-medium instruction setting in South Korea. Students’ narratives reveal how ChatGPT mediates between formal classroom expectations and informal, self-directed learning practices. The analysis highlights the ways AI use enables learners to assert agency—redefining what counts as legitimate knowledge, resisting institutional constraints, and experimenting with voice. The paper argues that these insights can inform educators seeking to understand student autonomy, AI literacy, and evolving teacher–student relationships in the era of generative AI. Implications include the need for pedagogies that embrace, rather than ban, AI as a contested but generative learning space.
Introduction
The first time I asked my students to reflect on their use of ChatGPT, I expected short, generic responses. Instead, I received thoughtful accounts—stories of using AI at to translate a stubborn sentence, of asking it to explain a theory in simpler terms, of testing how to say something politely in English before emailing a professor. Some students described ChatGPT as a safety net, others as a sparring partner, and a few as a risky but exciting shortcut. Across these narratives, one common thread emerged: ChatGPT was not simply a tool; it was a space in which students could try out ideas, rehearse voices, and negotiate what “good writing” meant to them.
In this way, ChatGPT became what Gutiérrez (2008) calls a third space—a hybrid zone where the discourses and norms of formal education intersect with learners’ lived practices, experiences, and languages. In this space, students draw on both the “official” literacy of the classroom and the “unofficial” literacy of their everyday digital lives. For some, this meant adapting AI’s suggestions to fit an assignment rubric; for others, it meant bending academic conventions to better express a personal perspective.
While studies like Stroud and Du (2025) have examined students’ acceptance of ChatGPT through the Technology Acceptance Model, focusing on perceived usefulness and behavioral intention, far less is known about how students integrate ChatGPT into their identity as learners, writers, and meaning-makers. A sociocultural perspective shifts the emphasis from mere adoption to agency—how learners position themselves in relation to AI and make choices about when, how, and why to use it.
This paper examines undergraduate students’ reflections on ChatGPT use in an English-medium instruction (EMI) course at a Korean university. Drawing on sociocultural theory (Vygotsky, 1978; Lewis, Enciso, & Moje, 2020) and third space theory (Gutiérrez, 2008), I explore the ways students employ ChatGPT as a mediating artifact to navigate the tensions between institutional academic expectations and their own literacy practices. Through qualitative content analysis of over 50 student responses, the study asks:
How do students describe their use of ChatGPT in relation to their learning, writing, and identity?
In what ways does ChatGPT function as a third space that enables or constrains student agency?
What can educators learn from student perspectives to inform AI-inclusive pedagogy?
By centering student voice, this study contributes to the growing body of research on AI in education, offering a lens for teachers to understand ChatGPT not merely as a technological innovation, but as a contested and generative literacy space.
Theoretical Framework
Sociocultural Theory and Mediation
This study draws on Sociocultural Theory (SCT) as articulated by Vygotsky (1978) and expanded in literacy research by scholars such as Lantolf (2006) and Lewis, Enciso, and Moje (2020). SCT positions learning as a socially mediated process in which cognitive development occurs through interaction with cultural tools, symbols, and others. Central to this view is the Zone of Proximal Development (ZPD)—the space between what a learner can do independently and what they can achieve with assistance.
In the contemporary classroom, digital tools like ChatGPT can function as mediating artifacts, shaping how learners access, practice, and perform literacy. Mediation is not simply about transferring information; it is about transforming how learners think, act, and position themselves in relation to knowledge and authority. When students interact with ChatGPT—asking questions, refining prompts, evaluating responses—they are not only receiving assistance, but also engaging in metacognitive and identity work.
Third Space Theory
Third Space Theory (Gutiérrez, 2008) provides a framework for understanding how students integrate ChatGPT into their learning. In educational contexts, the third space is a hybrid zone where “official” school discourses and “unofficial” everyday literacies intersect, creating opportunities for new meanings and identities to emerge. Rather than viewing these spaces as separate, third space theorists argue that learning is enriched when educators recognize and build on the interplay between them.
In the case of AI-assisted learning, ChatGPT can serve as a third space where institutional expectations meet personal learning strategies. Students draw on AI in ways that are shaped by both academic norms—such as formal English syntax, research conventions, and assignment structures—and their own communicative repertoires, including multilingualism, digital fluency, and non-linear problem-solving. This hybrid space allows for experimentation.
Student Agency in AI-Mediated Learning
Building on Lewis et al. (2020), this study conceptualizes student agency not simply as independent action, but as the ability to make purposeful, strategic choices within and against structural constraints. Agency involves deciding when to accept AI suggestions, when to modify them, and when to reject them entirely. It also includes critical engagement—questioning the accuracy, ethics, and implications of AI-generated content.
This view aligns with calls for critical literacy in education (Leander & Burriss, 2020), which emphasize the importance of equipping learners to navigate AI with discernment and intentionality. These acts of negotiation are central to the third space, where agency is enacted through the blending, remixing, or contesting of discourses.
In sum, this theoretical framing positions ChatGPT not as a neutral tool, but as a site of sociocultural mediation and third space activity, where students’ literacy practices are expanded, contested, and redefined. This perspective informs the methodological approach, which seeks to capture not only what students do with AI, but how they understand and position themselves in relation to it.
Methodology
Research Design
This study employs Qualitative Content Analysis (QCA) to examine undergraduate students’ reflections on their use of ChatGPT for academic work. QCA was chosen for its flexibility in identifying patterns, meanings, and themes within open-ended responses, allowing for both inductive and theory-driven coding (Lincoln & Guba 1985; Saldaña, 2021). Given the study’s grounding in Sociocultural Theory and Third Space Theory, the analysis prioritizes how students narrate their experiences, position themselves as learners, and describe interactions with AI as part of their literacy practices.
Context and Participants
Data were collected in an English-Medium Instruction (EMI) course at a university in South Korea. The course enrolled 50 undergraduate students from a range of majors, including education, social sciences, and business. While all students had at least intermediate English proficiency, their confidence and comfort with academic writing varied widely.
As part of a semester-end reflection activity, students were asked:
How have you used AI tools (such as ChatGPT) for your English language assignments this semester?
What did you learn about using AI from this experience?
Students could only respond in English as that was the required language for the class.
Data Collection
In total, 50 written reflections were collected. These responses ranged from short paragraphs to multi-page narratives. Students were informed that their reflections might be used for research purposes and that their names and identifying details would be anonymized.
Analytic Approach
Analysis proceeded in three stages:
Initial Reading and Open Coding
Responses were read multiple times to gain familiarity with the data.
Initial codes captured concrete actions (e.g., “used for grammar correction,” “checked theory definitions,” “translated sentences”) and expressed stances (e.g., “distrust in AI sources,” “AI as partner”).
Thematic Coding with Theory-Driven Categories
Codes were clustered into broader categories aligned with the theoretical framework:
Functional Use (e.g., grammar, vocabulary, structure)
Learning Enhancement (e.g., understanding theory, expanding perspectives)
Agency and Ethics (e.g., decision-making, originality, academic integrity)
Third Space Practices (e.g., blending academic and personal literacies)
Interpretive Analysis
Themes were interpreted in relation to Sociocultural Theory (mediation, ZPD) and Third Space Theory (hybrid literacies, identity negotiation).
Representative quotes were selected to illustrate each theme, prioritizing variety in voice and perspective.
Researcher Positionality
The researcher is both the course instructor and the analyst for this study. This dual role provided deep contextual knowledge of the course and student cohort, but also required reflexivity to minimize bias. With qualitative research, especially practitioner-based inquiry such as this, it is important to note my intersections, such as being a white middle-aged U.S. American university professor who has lived long term as a resident in South Korea.
Findings
Analysis of the 50 student reflections revealed four interrelated themes: (1) ChatGPT as a Functional Tool, (2) ChatGPT as a Learning Partner, (3) Agency and Ethical Positioning, and (4) Third Space Practices and Hybrid Literacies. Across these themes, students demonstrated nuanced decision-making, critical awareness, and a blending of academic and everyday literacies — all hallmarks of a Third Space where formal and informal practices intersect.
1. ChatGPT as a Functional Tool
For many students, ChatGPT’s most immediate value lay in its capacity to assist with grammar correction, vocabulary choice, and text organization. Students described using it to produce cleaner, more polished English while still maintaining ownership of their ideas.
“AI is very helpful in correcting grammar when doing tasks. AI is often used to refine complex sentences more naturally. In addition, various vocabulary and expressions can be learned through AI.”
“For example, I used AI to check the structure of my reflection essay and it helped me express my thoughts more clearly and politely.”
From a sociocultural perspective, these uses illustrate AI as a mediating artifact that helps students work within their Zone of Proximal Development — bridging the gap between their current proficiency and the desired academic standard.
2. ChatGPT as a Learning Partner
Several students framed ChatGPT not as a static tool but as a collaborator — a space for idea testing, clarifying complex theories, and generating examples. These students described an interactive process that mirrors collaborative dialogue in SCT.
“Through asking questions and getting feedback, I gain a deeper understanding of my English assignments.”
“I received a lot of help while working on English assignments using AI, especially when studying various theories… each time I was helped to double-check the definition of each theory and analyze the commonalities and differences.”
In these cases, ChatGPT functioned as a digital co-participant, supporting metacognitive growth and enabling iterative refinement of ideas before they were shared in class or submitted for assessment.
3. Agency and Ethical Positioning
While many students embraced ChatGPT, they also articulated limits, cautions, and ethical boundaries. Students frequently mentioned cross-checking information, resisting overreliance, and maintaining originality.
“Even if I used AI, I had to check the actual source separately because I wasn’t sure whether the source was accurate or not.”
“I try to make sure the ideas are still my own and that I’m learning, not just copying.”
These reflections reveal critical AI literacy — the ability to evaluate and filter machine-generated content — and position students as agents who negotiate the balance between efficiency and integrity. Within the Third Space framework, this is evidence of students actively shaping how institutional norms (academic honesty, critical thinking) interact with emerging AI literacies.
4. Third Space Practices and Hybrid Literacies
Perhaps the richest insights came from moments where students blended formal academic discourse with informal, personal, or multilingual practices. ChatGPT was used to translate Korean thoughts into English, adapt tone for cultural appropriateness, or rehearse ideas in informal language before “academicizing” them.
“First of all, I wanted to ask the professor a question during the assignment, but when it was difficult to express my opinion in English, I was able to send a message using the AI translation function to convey my intention more clearly.”
“I felt that AI was not something that would replace me, but a partner that would develop my thoughts with me.”
Here, ChatGPT operates as a third space — a zone where multilingual, culturally situated knowledge meets the demands of academic English. This hybridity allows students to retain elements of their personal voice and linguistic identity while also meeting institutional expectations.
Synthesis Across Themes
Across all four themes, students demonstrated that ChatGPT was not simply an “answer machine” but a dynamic literacy environment. Their reflections suggest that in AI-mediated spaces, agency is enacted through choice-making, critical evaluation, and hybrid language use. This aligns with Gutiérrez’s (2008) view of Third Space as a transformative zone — one that holds both the potential to reinforce institutional norms and to challenge or reshape them.
Discussion
The findings of this study position ChatGPT not as a neutral technology, but as a socially situated literacy space where students navigate between institutional demands and personal learning repertoires. Through the lens of Sociocultural Theory (Vygotsky, 1978; Lewis, Enciso, & Moje, 2020) and Third Space Theory (Gutiérrez, 2008), student reflections reveal how AI-mediated learning is both an extension of, and a departure from, traditional academic literacy practices.
ChatGPT as a Mediating Artifact in the ZPD
In SCT, mediation involves more than transferring knowledge; it transforms the nature of the task and the learner’s relationship to it. Students’ use of ChatGPT for grammar correction, vocabulary enhancement, and structural feedback reflects this mediational role. By enabling learners to perform at a higher linguistic level than they could independently, ChatGPT operates within their Zone of Proximal Development, scaffolding their journey toward more fluent, academically acceptable expression.
Third Space: Where Formal and Informal Literacies Meet
The data reveal that ChatGPT frequently functions as a third space—a hybrid zone in which academic writing conventions merge with students’ personal, multilingual, and digital literacy practices. This is evident in instances where students rehearsed ideas informally with AI before adapting them to formal assignments, or where they used AI translation features to communicate more effectively with their professor. These practices demonstrate how the boundaries between “school” and “non-school” literacies are porous, with AI acting as a bridge.
Importantly, this bridging does not simply transmit institutional norms into personal spaces; it also allows personal and cultural knowledge to travel into academic work. This mutual flow is what Gutiérrez (2008) identifies as the transformative potential of the third space.
Agency as Negotiation and Choice-Making
Student agency was evident in the ways learners made purposeful, strategic decisions about their engagement with ChatGPT. Many resisted overreliance, critically evaluated AI outputs, and emphasized maintaining authorship over their ideas. This aligns with Lewis et al.’s (2020) framing of agency as action within and against structural constraints.
What emerges is a portrait of students as active navigators—not passive consumers—of AI-generated content. They decide when AI serves their goals and when it might undermine them. This is a key insight for educators concerned that AI use erodes independent thinking: for many students, the opposite appears to be true.
Pedagogical Implications for Teachers
From these findings, several implications arise for educators:
Recognize AI as a Learning Space, Not Just a Tool
ChatGPT use is not limited to task completion; it is part of students’ meaning-making process. Treating it as such opens space for richer discussions about learning strategies and writing processes.
Integrate Critical AI Literacy into Instruction
Many students already engage in fact-checking and ethical reasoning. Explicit instruction in evaluating AI outputs can deepen these skills and prepare students for responsible, reflective AI use.
Leverage the Third Space for Engagement
Assignments that acknowledge students’ informal literacies (multilingual practices, personal voice, creative experimentation) can make academic work more authentic and empowering.
Shift from Policing to Accompaniment
Instead of focusing solely on detecting “AI misuse,” teachers can position themselves as partners who help students reflect on and refine their AI interactions. This fosters trust and promotes responsible agency.
Positioning This Study in the Literature
While Stroud and Du (2025) show that perceived usefulness predicts ChatGPT adoption, this study goes further by revealing how students integrate AI into their identity as learners. The qualitative, theory-driven approach offers a richer account of student perspectives, showing that adoption is not the endpoint—agency, negotiation, and hybridity are equally important.
Conclusion
This study explored how undergraduate students in an English-medium instruction course in South Korea use ChatGPT as a site of learning, identity negotiation, and agency. Through a qualitative content analysis of 50 student reflections, the research framed ChatGPT as a third space (Gutiérrez, 2008) in which institutional academic literacies intersect with students’ personal, multilingual, and digital practices. Drawing on Sociocultural Theory (Vygotsky, 1978; Lewis, Enciso, & Moje, 2020), the analysis revealed that students engage with ChatGPT not merely as a functional tool, but as a mediating artifact that supports metacognition, fosters creativity, and enables strategic choice-making.
The findings show that many students are already practicing forms of critical AI literacy—evaluating AI outputs, resisting overreliance, and maintaining authorship of their ideas. Far from eroding agency, for these students AI appears to have created space for experimentation, reflection, and hybrid literacy practices that blend formal and informal learning worlds. These insights suggest that educators who recognize and engage with AI as a learning environment can better support student growth and ethical engagement.
Limitations of this study include its single-institution context, reliance on self-reported data, and focus on one course. These factors may limit generalizability, though they also provide depth and contextual richness. Future research could expand to multiple institutions, disciplines, and cultural contexts, as well as explore longitudinal patterns in AI use over time.
In a rapidly evolving educational landscape, the question is not whether students will use AI, but how they will use it—and how educators will respond. By listening to students’ own accounts, we gain a more nuanced understanding of AI’s role in literacy learning. This study calls on educators to embrace the third space as a site of possibility: a place where institutional goals and students’ lived literacies meet, contest, and co-create the futures of learning.
References
Fullan, M. G. (1993). Why teachers must become change agents. Educational Leadership, 50(6), 12–17. https://www.ascd.org/el/articles/why-teachers-must-become-change-agents
Gutiérrez, K. D. (2008). Developing a sociocritical literacy in the third space. Reading Research Quarterly, 43(2), 148–164. https://doi.org/10.1598/RRQ.43.2.3
Lantolf, J. P. (2006). Sociocultural theory and L2: State of the art. Studies in second language acquisition, 28(1), 67-109.
Lantolf, J. P. (2006). Language emergence: Implications for applied linguistics—A sociocultural perspective. Applied linguistics, 27(4), 717-728.
Lantolf, J. P. (2007). Sociocultural source of thinking and its relevance for second language acquisition. Bilingualism: Language and Cognition, 10(1), 31-33.
Leander, K. M., & Burriss, S. K. (2020). Critical literacy for a posthuman world: When people read, and become, with machines. British Journal of Educational Technology, 51(4), 1262-1276.
Lewis, C., Enciso, P. E., & Moje, E. B. (Eds.). (2020). Reframing sociocultural research on literacy: Identity, agency, and power. Routledge. https://doi.org/10.4324/9780429448317
Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry. SAGE.
Saldaña, J. (2021). The coding manual for qualitative researchers (4th ed.). SAGE.
Stroud, R., & Du, J. (2025). Student acceptance of ChatGPT in higher education: A mixed method study of attitudes. Journal of Applied Learning & Teaching, 8(2).
Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes (M. Cole, V. John-Steiner, S. Scribner, & E. Souberman, Eds. & Trans.). Harvard University Press.
Comments
Post a Comment