The research behind Second Listen.
Second Listen is built on research about reflection, assessment for learning, classroom talk, how pupils learn, coaching and automated feedback on teaching. Every source is labelled with the kind of evidence it is, because a magazine article, a meta-analysis and a randomised trial do not carry the same weight.
Reflective practice and reflective cycles
Second Listen's loop is the What? So what? Now what? model: describe what happened, make sense of why it matters, then plan one concrete action. The wider literature agrees on one point the product is built around: reflection changes practice only when it is specific, grounded in evidence and revisited, not vague or one-off.
| Source | Kind | The idea |
|---|---|---|
| Dewey (1933), How We Think | Foundational theory | Reflective thinking starts from a real problem in experience and tests ideas against evidence, not habit. |
| Schön (1983), The Reflective Practitioner | Foundational theory | Professionals learn by reflection-on-action: stepping back after the event to reframe what happened. |
| Kolb (1984), Experiential Learning | Theory | A cycle of experience, reflection, conceptualisation and active experimentation. |
| Gibbs (1988), Learning by Doing | Practice model | A six-stage cycle: description, feelings, evaluation, analysis, conclusion, action plan. |
| Borton (1970); Driscoll (1994); Rolfe, Freshwater and Jasper (2001) | Practice model | What? (the facts), So what? (what it means, and what knowledge helps), Now what? (an action plan with concrete steps). |
| University of Edinburgh Reflection Toolkit | Practitioner guide | Presents the same model with prompts for each stage, and stresses specific rather than vague action plans. |
| Brookfield (1995, 2nd ed. 2017), Becoming a Critically Reflective Teacher | Practitioner theory | Four lenses on teaching: your own view, pupils' view, colleagues' view, and theory and research. |
| Larrivee (2000), Reflective Practice 1(3); Loughran (2002), Journal of Teacher Education 53(1) | Peer-reviewed papers | Reflection needs a real puzzle, evidence and a change in action, or it becomes description. |
Assessment for learning and feedback
Three of the eight indicators sit on assessment for learning. The evidence is strong on direction and weaker on size, and several sources warn that feedback can make things worse.
| Source | Kind | The idea |
|---|---|---|
| Sadler (1989), Instructional Science 18 | Conceptual framework | Pupils improve when they know the standard, where they are against it, and how to close the gap. |
| Black and Wiliam (1998), Phi Delta Kappan 80(2) | Synthesis | Formative assessment raises standards when teachers use evidence of learning to adapt their teaching. |
| Leahy, Lyon, Thompson and Wiliam (2005), Educational Leadership 63(3) | Synthesis | Formative assessment as minute-by-minute, day-by-day classroom practice. |
| Hattie and Timperley (2007), Review of Educational Research 77(1) | Synthesis | Feedback answers three questions (where am I going, how am I going, where to next) at four levels; feedback about the self helps least. |
| Kluger and DeNisi (1996), Psychological Bulletin 119(2) | Meta-analysis | Across 607 effect sizes feedback helped on average, but over a third of feedback interventions lowered performance. |
| EEF Feedback strand and guidance report (2021) | Evidence summary | Feedback is high impact on average, and can have negative effects when done poorly. |
| Clarke (2014), Outstanding Formative Assessment | Synthesis | Turns Sadler and Black and Wiliam into everyday classroom method. |
| Double, McGrane and Hopfenbeck (2020), Educational Psychology Review 32(2) | Meta-analysis | Peer assessment helps learning; peer grading does not. |
| Zimmerman (2002); Nicol and Macfarlane-Dick (2006) | Conceptual frameworks | Self-regulated learning: pupils plan, monitor and evaluate their own learning. |
| EEF Metacognition and Self-Regulated Learning | Evidence summary | Teaching pupils to plan, monitor and evaluate is high impact for low cost. |
| EEF Mastery learning | Evidence summary | Secure understanding before moving on. |
Classroom talk, questioning and dialogue
Talk is what a microphone captures best, so this research does the most work in a reflection. The consistent finding is that how a teacher asks, waits and builds on answers shapes how much pupils think, though most of the evidence is correlational or small.
| Source | Kind | The idea |
|---|---|---|
| Rowe (1986), Journal of Teacher Education 37(1) | Peer-reviewed study | Lengthening the pause after a question from about one second to three or more gives longer answers, more volunteers and more reasoning. |
| Nystrand and Gamoran (1991), Research in the Teaching of English 25(3) | Peer-reviewed study | Authentic questions, and uptake of what pupils say, mark discussion rather than recitation. |
| Mercer, Wegerif and Dawes (1999), British Educational Research Journal 25(1) | Peer-reviewed study | Teaching ground rules for exploratory talk changed how pupils reasoned together. |
| EEF Dialogic Teaching trial (2017) | Randomised trial | Across 76 schools, dialogic teaching gave small gains in English, science and maths in Year 5. |
| Howe, Hennessy, Mercer, Vrikki and Wheatley (2019), Journal of the Learning Sciences 28(4-5) | Peer-reviewed study | Elaboration and querying predicted attainment where pupil participation was high. |
| O'Connor, Michaels, Chapin and Harbaugh (2017), Learning and Instruction 48 | Peer-reviewed study | Whole-class discussion helped learning, but how much an individual pupil spoke did not predict their own outcome. |
| EEF Oral language interventions | Evidence summary | Structured talk improves language and reasoning. |
| EEF Collaborative learning; EEF Peer tutoring | Evidence summary | Structured group work and peer tutoring help, especially for consolidating learning. |
How pupils learn: instruction, practice and creativity
Research from cognitive science and on creative and play-based teaching answers the moments that talk research does not: how new material is introduced, practised and remembered.
| Source | Kind | The idea |
|---|---|---|
| Rosenshine (2012), American Educator 36(1) | Synthesis | Principles of instruction: small steps, a high success rate, and checking understanding. |
| Roediger and Karpicke (2006), Psychological Science 17(3) | Peer-reviewed study | Recalling material strengthens long-term memory more than rereading it. |
| Dunlosky et al. (2013), Psychological Science in the Public Interest 14(1) | Synthesis | Practice testing and spacing are the most useful study techniques; self-explanation and elaborative questioning help moderately. |
| Sweller and Cooper (1985), Cognition and Instruction 2(1) | Peer-reviewed study | Studying worked examples beats solving equivalent problems for novices. |
| Wood, Bruner and Ross (1976), Journal of Child Psychology and Psychiatry 17(2) | Peer-reviewed study | Scaffolding: support that is withdrawn as the learner gains control. |
| Alfieri et al. (2011), Journal of Educational Psychology 103(1) | Meta-analysis | Unassisted discovery loses to explicit teaching; guided discovery wins. |
| Lee et al. (2015), Review of Educational Research 85(1) | Meta-analysis | Drama-based teaching gives modest gains in achievement. |
| Skene et al. (2022), Child Development 93(4) | Meta-analysis | Guided play beats direct instruction on some early maths and spatial outcomes. |
Coaching and teacher professional learning
Second Listen works like a coach's ear between coaching conversations: it brings evidence from the lesson, asks a question, and helps the teacher commit to one specific change. The professional learning evidence backs that shape, and warns that effects shrink as programmes grow.
| Source | Kind | The idea |
|---|---|---|
| Kraft, Blazar and Hogan (2018), Review of Educational Research 88(4) | Meta-analysis | Across 60 causal studies, coaching improved teachers' instruction and pupils' achievement. |
| Sims, Fletcher-Wood et al. (2021), EEF systematic review; EEF Effective Professional Development guidance (2021) | Systematic review and guidance | Fourteen mechanisms in four groups make professional development work: build knowledge, motivate teachers, develop teaching techniques, embed practice. |
| Gollwitzer and Sheeran (2006), Advances in Experimental Social Psychology 38 | Meta-analysis | Across 94 tests, if-then plans (when X happens, I will do Y) helped people reach their goals. |
| Sheeran, Listrom and Gollwitzer (2024), European Review of Social Psychology 36(1) | Meta-analysis | Across 642 tests, plans worked best in an if-then format, with high motivation and when rehearsed. |
| Timperley et al. (2007), Teacher Professional Learning and Development: Best Evidence Synthesis | Synthesis | Professional learning works when teachers inquire into their pupils' learning, try something, and check its effect. |
| Knight (2018), The Impact Cycle | Practitioner model | Instructional coaching as identify, learn, improve, starting from a clear picture of reality such as a recorded lesson. |
Recorded lessons, AI and automated feedback
| Source | Kind | The idea |
|---|---|---|
| Tripp and Rich (2012), Teaching and Teacher Education 28(5) | Review and small qualitative study | Teachers said reviewing their own lessons helped them focus, see their teaching afresh, take responsibility for improving and see their own progress. |
| Kelly, Olney, Donnelly, Nystrand and D'Mello (2018), Educational Researcher | Peer-reviewed study | Speech recognition and machine learning picked out authentic questions in classroom audio, broadly agreeing with human coders. |
| Demszky, Liu, Hill, Jurafsky and Piech (2023), Educational Evaluation and Policy Analysis | Randomised trial | Private, automated feedback on how instructors took up students' ideas raised that uptake by 13% across 1,136 instructors. |
- Quotes, not scores. A reflection returns moments quoted from the lesson, each with a question. Nothing is graded, and nothing about a teacher's own lesson is shown to anyone else.
- The AI never writes a citation. It chooses an approach from a bank of 29 checked sources. The reference and its caveat are added afterwards, so a made-up reference cannot appear.
- Audio only, deleted quickly. No video. The audio is deleted within seconds of being transcribed, and the quoted moments after 14 days (up to six weeks if the teacher keeps working on that lesson), because reflection needs the moments, not the recording.
Want the full reference list?
We’re happy to share the complete references, and the caveat we show teachers for each source.
Ask for the references