For some time now, I had been working on a chapter about reflective writing in language teacher education, for an edited volume on that topic. That text went in a different direction, but I thought that some of the ideas that did not fit there may be useful here, with most of the scholarly apparatus left out. The draft I abandoned stemmed from my frustration with some aspects of reflective writing in my teaching, and I think that both the genre and the challenge will be familiar to many readers. The pedagogical rationale for such tasks is difficult to defend when these became routinised (Akbari, 2007; Mann & Walsh, 2013); it has become even harder now, as generative AI can produce plausible reflective prose on demand.
The argument I developed in the draft, which I summarise in this post, is that the question is less new than it looks, and that answering it involves revisiting what we think of as reflective writing. Typically, we view this as a trace of student reflection, but I think it may be more productive to treat it as a place where reflection happens. Reflective writing, in this perspective, is theorising in real time, or ‘writing-assisted reflection’.
A record of reflection, or a space for it?
For most of us involved in teacher education, reflective writing is a loose genre that includes things like journals, portfolios and reflective essays, which students usually submit towards the end of an initial teacher education or professional development module, and which we typically use to assess how much learning has taken place. The challenge we are now facing is that such texts are often produced by artificial intelligence, and — some say — are no longer useful or relevant. If a chatbot can write the journal entry, the argument goes, what is he journal entry for?1
Reflection as disciplined inquiry
We generally tend to think of reflection as a kind of introspection associated with teaching, but its original function in some foundational educational thinking was that of a disciplined mode of inquiry. Dewey (1933), for instance, conceptualised reflection as a process that began with a moment of uncertainty, when routine action became unfamiliar, and proceeded as a deliberate, evidence-based attempt to resolve that uncertainty. Thinking along similar lines, Schön (1983) suggested reflection (in- and on-action) as ways to surface knowledge that professionals cannot yet fully articulate, and to make it available for scrutiny.
Despite differences in outlook, what these two accounts share is a view that reflection is an act of knowledge production (or, to be technical, ‘epistemic work’) leading to situated understandings that the practitioner did not have before. Following Edge (2011), I refer to this work as theorising2 (let’s keep this term in mind, as it will appear many times in what follows). Notably, writing does not feature prominently in either account.
The role of writing in teacher reflection
The reflective writing task is a somewhat more recent addition to the repertoire of teacher education. It became more prominent as reflective models of teacher education became more widespread, the idea being that this was the most efficient way to encourage, observe, and (crucially) assess reflection (Farrell, 2016). There’s nothing wrong with any of this. So far.
But here’s where things get complicated: Because reflective writing usually takes place in assessed programmes, it is increasingly scaffolded by criteria that help teachers understand what a ‘good’ reflection looks like. But criteria that reward the appearance of reflection can have the unintended consequence of shifting attention away from theorising (the epistemic process I mentioned above) and towards meeting formal conventions. When that happens, writing risks becoming a performance of reflection rather than a space for it.
Doubts of this kind predate artificial intelligence by decades. Hatton and Smith (1995) found that many reflective writing tasks tend to produce descriptive texts with little sign of reflection. A decade later, Hobbs (2007) pointed out that, when teachers engage in mandated reflective writing, they are quite capable of turning in texts that meet all the formal requirements, but do not involve any actual reflection. Similarly, Mann and Walsh (2013) note that reflective practice has become quite ritualised, and that the texts that students produce are generally quite formulaic.
Thought generation vs. thought representation
One important distinction we can draw, then, is between writing as thought representation and writing as thought generation. This is similar to a point that Richardson (2000) makes in the context of qualitative research: she describes writing as a method of inquiry, which helps researchers make sense of what they think. Somewhat relatedly, Johnson and Golombek (2011) suggest that when teachers produce autobiographical narratives, they externalise their developing understandings, verbalise them, and interrogate them systematically. Seen this way, reflection is an opportunity for theorising, writing can mediate it, and the reflective text is the trace that the process leaves behind. This is something I explored, from a more practical angle, in an earlier post on AI-assisted teacher reflection.
Theorising is epistemic work (i.e., knowledge-building) through which teachers build, test, and revise tentative explanations about their professional context and their practice in it.
The trouble starts when we only assess the trace. Course participants may then learn to produce texts that display the expected features of reflection (e.g., performative self-questioning or predictable ‘learning outcomes’), without having done the intellectual work that those features are meant to represent. Generative AI has pushed this risk to its extreme, because it makes it possible to decouple the trace from the teacher’s reflection altogether.
What writing makes possible
And all this brings us back to the question: why is reflective writing still relevant to teacher education? I think that answering this involves identifying what the process of composing a text makes possible for reflection, which is not available (or at least not equally available) when reflection is confined to thought or talk. I think there are four such affordances. None of them is exclusive to writing, but all of them are harder to come by without it.
| Affordance | How it supports reflection | What teachers do |
|---|---|---|
| Externalisation | Makes reflection visible | Explicitly articulate assumptions and interpretations |
| Permanence | Makes reflection revisable | Re-read, compare and revise earlier reflections |
| Ordering | Makes reflection coherent | Tentatively reconstruct temporal sequences and causal relationships |
| Addressability | Makes reflection shareable | Engage with questions and alternative interpretations from others |
Externalisation
Much of the knowledge that teachers use in action is tacit, and what writing does is articulate it into something observable. A vague intuition, such as “I find classroom management hard”, becomes a specific claim (“Today, students became disruptive when…”), perhaps followed by tentative explanations. Seeing one’s developing knowledge from this distance makes self-questioning easier, and it helps to notice gaps and contradictions that may not be visible otherwise.
Permanence
Scripta manent, and all that: written reflections endure as objects for us to revisit (Ong, 1982). They are — to draw on intentionality theory — sedimented meaning. In addition to providing a record documenting the emergence of understanding (Farrell, 2013), this creates the opportunity to revisit these developing understandings, when they know more or have more data, and to keep revising them.
Ordering
Classroom experience is simultaneous and messy. Writing imposes structure onto it,3 either in the form of linear sequences (“first…, then…”) in the case of narratives, or in the form of causal connections (“because…, therefore…”) in explanations. These ordering moves are powerful heuristics, and they are, largely, how writers often discover what they think while they are writing.
Addressability
Learning is social, and writing helps to bring this out. Written reflections are artefacts that other teachers (including future cohorts or teachers in other places) can read, compare with their experience, and react to. They can become shared objects for collaborative reflection (Mann & Walsh, 2013), and the ensuing dialogue can draw on a broad pool of professional experience to extend the teacher’s original thinking.
What generative AI changes
Whether AI supports or undermines reflection seems to depend less on whether it is used (as will almost inevitably happen) than on where it enters the writing process. I am still working through the implications, but three sets of them seem to follow.
For teacher educators
What the simple typology of affordances above (externalisation, permanence, ordering, addressability) offers is a scaffold for designing reflective writing tasks around what writing makes possible.
In practical terms, this could mean activities like:
- creating writing prompts that rely on specific classroom evidence (rather than generic reflection);
- designing tasks that involve sequences of drafts (rather than single submissions);
- using AI to generate competing explanations or counterfactual scenaria (“what would have happened if…”) and using these for reflection;
- Where AI is concerned, it means modelling responsive uses (“ask the tool to challenge your explanation”), explicitly discouraging generative ones (“ask the tool to write your reflection”), and building reflection on AI use into the tasks themselves.
For programme and assessment designers
The most important (and arguably overdue) change that AI has made necessary is a shift from the trace of the reflection (i.e., the text) to the process of reflection. In practical terms, this could involve grading criteria that reward revision, reasoning, consideration of alternatives and responsiveness to feedback. It could also mean that at least some long-standing assessment genres, such as the single end-of-practicum reflective essay, may need rethinking (or retirement).
I am also very sceptical about the effectiveness of blanket bans on the use of AI.4 I believe it might be more effective to focus on the difference between responding and replacing, and to demand transparency about how participants used AI.
For practising teachers
One change I would like to see happen is if practising teachers could reclaim reflective writing as a tool for thinking rather than an obligation. This involves writing first, and then asking for challenge: using AI as a sparring partner, not a ghostwriter, and treating its suggestions as hypotheses to test against one’s own classroom, not as conclusions.
It also involves keeping reflective writing as a cumulative record that can be revisited, so that one’s own development becomes visible. Written reflection that moves from description to explanation and evidence is, moreover, a first step towards teacher inquiry (Borg, 2010), which is where reflective practice begins to connect with research literacy.
A closing thought
The point I have been trying to make is that generative AI has not made reflective writing obsolete. What it has done is make clearer what reflective writing was for. It confronts us with the fact that practices which only collect texts are pedagogically weak, and it opens the possibility for practices designed around externalisation, permanence, ordering and addressability, that is, practices that can absorb AI, and perhaps even benefit from what it makes possible.
References
Click to view the reference list
Akbari, R. (2007). Reflections on reflection: A critical appraisal of reflective practices in L2 teacher education. System, 35(2), 192–207. https://doi.org/10.1016/j.system.2006.12.008
Borg, S. (2010). Language teacher research engagement. Language Teaching, 43(4), 391–429. https://doi.org/10.1017/S0261444810000170
Dewey, J. (1933). How we think: A restatement of the relation of reflective thinking to the educative process. D. C. Heath.
Edge, J. (2011). The reflexive teacher educator in TESOL: Roots and wings. Routledge.
Farrell, T. S. C. (2013). Teacher self-awareness through journal writing. Reflective Practice, 14(4), 465–471. https://doi.org/10.1080/14623943.2013.806300
Farrell, T. S. C. (2016). Anniversary article: The practices of encouraging TESOL teachers to engage in reflective practice: An appraisal of recent research contributions. Language Teaching Research, 20(2), 223–247. https://doi.org/10.1177/1362168815617335
Hatton, N., & Smith, D. (1995). Reflection in teacher education: Towards definition and implementation. Teaching and Teacher Education, 11(1), 33–49. https://doi.org/10.1016/0742-051X(94)00012-U
Hobbs, V. (2007). Faking it or hating it: Can reflective practice be forced? Reflective Practice, 8(3), 405–417. https://doi.org/10.1080/14623940701425063
Johnson, K. E., & Golombek, P. R. (2011). The transformative power of narrative in second language teacher education. TESOL Quarterly, 45(3), 486–509. https://doi.org/10.5054/tq.2011.256797
Mann, S., & Walsh, S. (2013). RP or ‘RIP’: A critical perspective on reflective practice. Applied Linguistics Review, 4(2), 291–315. https://doi.org/10.1515/applirev-2013-0013
Ong, W. J. (2012). Orality and literacy: The technologizing of the word (30th anniversary ed.). Methuen. (Original work published 1982)
Richardson, L. (2000). Writing: A method of inquiry. In N. K. Denzin & Y. S. Lincoln (Eds.), Handbook of qualitative research (2nd ed., pp. 923–948). SAGE.
Schön, D. A. (1983). The reflective practitioner: How professionals think in action. Basic Books.
By subscribing to this blog, you will receive occasional updates on language education, including my ongoing work on AI in language teaching and learning and on language teachers’ research literacy. (privacy policy)
Notes
- When this question was last asked in a meeting in an institution I would rather not name, I countered that no parking signs in Greece are often disregarded, so — by analogy — they should be taken down. I was advised that I was being unhelpful. ↩︎
- Edge’s phrase for this is the teacher’s ‘best understanding of what is actually going on’ (2011, p. 80). ↩︎
- Unlike art, which is about imposing chaos onto order. I think that was Adorno who said that. ↩︎
- I am old enough to remember bans on using calculators at school, and remember English teachers who would not let us use dictionaries. ↩︎
Modernisation for whom? Reading Greece’s foreign-language reform
A short report presents Greece’s foreign language reform as a “system-wide modernisation”. Read closely, the reform speaks the language of multilingualism while organising itself around English, pilots AI before governing it, and leaves open questions about evidence, materials, and teacher readiness.
Languages unite, but will AI make them more equal?
For the 2026 European Day of Languages, I explore what “Languages unite!” means when our linguistic choices are increasingly mediated by AI technology.
AI-Assisted Teacher Reflection: Thinking Before, During, and After the Lesson
Teaching has always required reflective practice. This post explores how artificial intelligence can support three types of teacher reflection (before, during, and after the lesson) drawing on Farrell’s framework and the AI Lang guidelines.
Frequently asked questions
Has generative AI made reflective writing pointless?
No. It has made it harder to justify tasks that only collect reflective texts, because such texts can now be generated on demand. The value of reflective writing lies in what the act of composing makes possible for the teacher’s thinking, and that has not changed.
What does writing add to reflection that thinking or talking does not?
Four things, none of which is exclusive to writing but all of which are harder to come by without it: externalisation (thought becomes visible), permanence (it can be revisited and revised), ordering (events are put in sequence and connected causally), and addressability (others can read and respond).
Can teachers use AI in their reflective writing?
It depends on where AI enters the process. When it composes the text in the teacher’s place, it bypasses reflection. When it responds to a text the teacher has written, for instance by challenging an explanation or surfacing an assumption, it may support it. Compose first, then ask for challenge.
How can reflective writing be assessed when AI can generate it?
By shifting attention from the finished text to the process it encodes: change between drafts, grounding in classroom evidence, consideration of alternative explanations, and responsiveness to feedback. None of this guarantees that reflection has taken place, but it rewards engagement rather than fluency in the genre.
Summary
- Reflective writing is usually treated as a record of reflection. It is more productive to treat it as a space where reflection happens.
- Composing a text affords externalisation, permanence, ordering and addressability. These are what is lost when composition is offloaded.
- Generative AI did not create the gap between reflective text and reflection. It removed the effort that used to disguise it.
- Policing who wrote the text protects the product. Redesigning tasks and assessment around the process is the more useful response.
- AI can support reflection when it responds to what teachers write, and undermines it when it writes in their place.
Additional reading
- A more practical companion to this post: AI-Assisted Teacher Reflection: Thinking Before, During, and After the Lesson.
- Mann, S., & Walsh, S. (2013). RP or ‘RIP’: A critical perspective on reflective practice. Applied Linguistics Review, 4(2), 291–315.
- Hobbs, V. (2007). Faking it or hating it: Can reflective practice be forced? Reflective Practice, 8(3), 405–417.
- Johnson, K. E., & Golombek, P. R. (2011). The transformative power of narrative in second language teacher education. TESOL Quarterly, 45(3), 486–509.
- Farrell, T. S. C. (2016). The practices of encouraging TESOL teachers to engage in reflective practice. Language Teaching Research, 20(2), 223–247.

About me
Achilleas Kostoulas is an applied linguist and language teacher educator at the Department of Primary Education, University of Thessaly, Greece. He holds a PhD and an MA in Teaching English to Speakers of Other Languages from the University of Manchester, UK and a BA in English Studies from the National and Kapodistrian University of Athens, Greece.
His research explores a wide range of issues connected with language (teacher) education, including language contact and plurilingualism, linguistic identities and ideologies, language policy and didactics, often using a Complex Dynamic Systems Theory perspective to tease out connections between them. Some of his work includes the research monograph The Intentional Dynamics of TESOL (2021, De Gruyter; with Juup Stelma) and the edited volume Doctoral Study and Getting Published (2025, Emerald; with Richard Fay), as well as numerous other publications.
Achilleas currently contributes to several projects that bring together his long-standing interests in language education, teacher development, and the social dimensions of language learning. As the coordinator of the AI Lang (Artificial Intelligence in Language Education) expert team at the European Centre for Modern Languages of the Council of Europe, he is active in developing principles and resources to help educators make informed, pedagogically grounded use of AI in their teaching. He also leads the University of Thessaly team of ReaLiTea (Research Literacy of Teachers), a project that supports language teachers to engage with, and contribute to, educational research. Alongside these, he contributes to LocalLing (Revitalisation of Linguistic Diversity and Cultural Heritage), a Horizon-funded initiative for the preservation of heritage and minority languages globally.
In addition to the above, Achilleas is the (co)editor-in-chief of the European Journal of Education and Language Review, and welcomes contributions that explore the dynamic intersections between language, education, and society.
About this post
This blog is a space for slow, reflective thinking about applied linguistics, language education, professional development, and the role of technology in language teaching and learning. Transparency about process, tools, and authorship is part of that commitment.
- I wrote this post on 15 November 2026. It summarises the argument of a chapter that I was preparing for an edited volume on reflective writing in language teacher education, but eventually decided to take to a different direction. I will periodically revise it to ensure accuracy, so feel free to point out any issues that come to your attention.
- The post is based on my own chapter draft. I used artificial intelligence to support copy-editing and Search Engine Optimisation. I retain responsibility for analytical thinking, authorial decisions and wording.
- The views expressed here are personal and do not necessarily reflect those of the University of Thessaly, or any other entity with which I am affiliated.
- The featured image, by Matt Fowler KC, is licenced from Adobe Stock.



Leave a Reply