Abstract

This practitioner enquiry examines how the think-aloud strategy, as a metacognitive instructional tool, supports high-quality teaching and learning among three Chinese ESL learners aged 6 - 8 in a private learning centre in China. Drawing on structured research log observations and a child-adapted visual questionnaire across two consecutive sessions, the study applies reflexive thematic analysis to identify patterns in metacognitive engagement, self-monitoring, strategy use, and affective participation. Findings indicate that the strategy promoted metacognitive awareness across all proficiency groups, with the most substantial gains in the lower-proficiency learner. Questionnaire data corroborated observation findings and revealed a consistent shift in learner enjoyment and willingness to re-engage. Structured think-aloud instruction is concluded to be pedagogically promising in early ESL settings when embedded within responsive and emotionally supportive teaching practice.

Keywords

Think-Aloud Strategy, Metacognition, Self-Regulated Learning, ESL, Young Learners, Practitioner Enquiry, Second Language Acquisition, Scaffolding, Chinese Learners, High-Quality Teaching

Share and Cite:

1. Introduction

What prompted this enquiry was a pattern I noticed repeatedly in my own classroom: learners who appeared to understand a task would arrive at incorrect answers or complete tasks silently in ways that made it impossible to identify where their reasoning had broken down. This observation pointed toward a gap between surface comprehension and the deeper self-awareness that effective learning requires. Teaching that addresses this gap draws learners into active awareness of their own thinking rather than simply transmitting knowledge, and reflects what the Education Endowment Foundation (2018) identifies as high-impact pedagogical practice. Within this framework, metacognitive strategies have attracted sustained research interest for their potential to develop learners’ capacity to think ahead about how to approach a task, check their understanding as they work through it, and reflect on how well they did afterwards (Flavell, 1979; Schraw & Dennison, 1994). Metacognition has been shown to correlate positively with improved learning outcomes across educational contexts (Perry et al., 2019).

In the field of second language acquisition (SLA), metacognitive awareness is particularly significant because language learners must simultaneously manage vocabulary, grammar, and communicative production (Cohen, 2005). When a child falls silent mid-task, it is rarely clear whether the difficulty lies in understanding the vocabulary, constructing the grammar, or simply not knowing where to begin. The think-aloud strategy asks learners to speak their thinking aloud as they complete a task, making that otherwise hidden process audible to both learner and teacher. Ericsson and Simon (1993) demonstrated that this kind of concurrent verbal output provides a valid and reliable record of cognitive processing, opening genuine opportunities for reflection, feedback, and instructional adjustment.

This enquiry was conducted in a private English learning centre in China, where three learners aged 6 - 8 participated in two structured lessons incorporating think-aloud activities. Commercial learning centres in China frequently operate within performance-oriented curricula where efficiency is prioritised over reflective learning (Hu, 2002), and cultural norms may discourage learners from expressing uncertainty aloud (Vandergrift & Goh, 2012). This context makes the investigation of metacognitive strategies both complex and pedagogically valuable.

The main question this study seeks to answer is: What is the role of the think-aloud strategy in facilitating metacognitive awareness and high-quality teaching and learning among Chinese ESL learners aged 6 - 8 in a private learning centre setting? The enquiry is structured around a critical literature review, an account of the research design and methodology, a presentation and discussion of findings, and a concluding section that draws out implications for practice.

For the purposes of this enquiry, high-quality teaching and learning is defined operationally rather than in the abstract: it is teaching that makes learners’ thinking visible and responsive, and learning that is accompanied by growing self-awareness, willingness to engage, and improved task accuracy. This definition is operationalised directly through the study’s two instruments. In the structured research log, it corresponds most closely to Domain 1 (Verbal Metacognitive Engagement), Domain 2 (Self-Monitoring and Self-Regulation), and Domain 4 (Affective and Social Indicators), which together capture whether learners are thinking aloud, catching and correcting their own errors, and engaging willingly rather than reluctantly. In the Child-Adapted Learner Questionnaire, it corresponds to Q2 (perceived helpfulness for understanding), Q5 (perceived helpfulness for error-finding), and Q6 (willingness to re-engage), which capture the same construct from the learner’s own perspective. Task accuracy, recorded independently through the Look and Write and Read and Write scores discussed in Section 3.2, provides a further, non-self-reported indicator. Claims made in Sections 4 and 5 about high-quality teaching and learning should be read against this specific, triangulated evidence base rather than as a claim about general attainment or curriculum coverage.

2. Literature Review

2.1. Metacognition, Self-Regulation, and Learning

The concept of metacognition was first theorised by Flavell (1979, p. 906) as “knowledge and cognition about cognitive phenomena”. This formulation captures a person’s capacity to be aware of and think about their own mental processes, including knowing what they find difficult, recognising when understanding has broken down, and deliberately choosing strategies to support their learning. It has since become central to educational research, with substantial evidence linking metacognitive awareness to academic achievement across age groups and subject areas (Veenman & Afflerbach, 2006). The Education Endowment Foundation (2018) identifies explicit metacognitive instruction as one of the highest-impact pedagogical approaches available to practitioners, a finding reinforced by Perry et al. (2019) across more than fifty studies.

Zimmerman (2002) extended metacognition into a broader model of self-regulated learning, arguing that effective learners take active control of their own cognitive behaviour before, during, and after tasks rather than passively receiving instruction. For younger learners, however, these capacities are still developing. Whitebread et al. (2009) found that children aged 5 - 7 demonstrate emergent metacognitive behaviours but require explicit instructional support to access and sustain them. Dent and Koenka (2016) found through a meta-analysis that metacognitive strategy instruction consistently improved outcomes across primary-age learners, with effect sizes increasing when instruction was embedded in authentic classroom tasks rather than isolated exercises.

2.2. Metacognitive Strategies in Second Language Acquisition

In SLA research, metacognitive strategies occupy a well-established position as facilitators of language monitoring and production. Cohen (2005) studied what distinguishes more successful language learners from less successful ones, finding that stronger learners are not simply more capable linguistically; they are more strategic. They think ahead about how to approach a language task, notice when communication is breaking down or when an error has occurred, and adjust their language use accordingly. Less successful learners tend to proceed without this self-monitoring. Cohen’s central argument is that these strategic behaviours can be taught to learners who do not develop them independently. Similarly, Vandergrift and Goh (2012) demonstrate that metacognitive awareness supports second language listening comprehension by enabling learners to check their understanding against contextual information and adjust interpretation accordingly. Teng (2020) further demonstrates that metacognitive knowledge and regulation play a significant mediating role in EFL learner performance, with learners deploying metacognitive strategies showing stronger monitoring and revision behaviours than those receiving conventional instruction.

An important distinction exists between metacognitive capacity and metacognitive deployment. Learners may possess an awareness of their thinking processes without spontaneously applying regulatory strategies during tasks (Schraw & Dennison, 1994). This gap is particularly notable in younger learners, where the executive function systems underpinning strategic self-regulation are still maturing. Accordingly, instructional scaffolding plays a critical role in bridging this gap for young ESL learners.

A further consideration concerns the relationship between metacognition and learner motivation in SLA contexts. Oxford (2017) argues that when a learner successfully uses a strategy to regulate their own learning and it works, they feel more capable, and this experience of success strengthens their belief that they can manage future challenges, motivating continued strategy use. For young ESL learners who may not yet see themselves as capable English speakers, even a single positive experience with think-aloud can begin to shift that self-perception. It is worth noting, however, that the evidence base for metacognitive strategy instruction in ESL settings with learners below the age of eight remains comparatively thin. Most studies cited in reviews such as Hajar and Karakus (2025), whose bibliometric analysis spans fifty years of language learning strategy research, focus on adolescent or adult learners, and findings may not translate straightforwardly to younger populations. This gap makes small-scale practitioner enquiries such as the present study particularly valuable in beginning to generate contextually grounded evidence.

2.3. The Think-Aloud Strategy: Theoretical Foundations

The think-aloud strategy draws theoretical legitimacy from the protocol analysis tradition established by Ericsson and Simon (1993), who demonstrated that concurrent verbalisation of thought provides a reliable window into cognitive processing. When used pedagogically, think-aloud allows learners to externalise their reasoning, reinforcing metacognitive monitoring and helping them identify errors or gaps in understanding. Fisher and Frey (2018) argue that teacher modelling of think-aloud provides learners with a visible example of expert strategic thinking, which they can subsequently internalise and apply independently. This approach is grounded in Vygotsky’s (Vygotsky, 1978) concept of the Zone of Proximal Development, which holds that the most productive instruction occurs in the gap between what a learner can do independently and what they can achieve with the support of someone more knowledgeable. Teacher-modelled think-aloud sits squarely in this framework: the teacher temporarily performs the metacognitive verbalisation the learner cannot yet manage independently, creating the conditions for the learner to internalise it over time. Hajar and Karakus (2025), in a bibliometric review spanning fifty years of language learning strategy research, highlight a renewed interest in think-aloud methods as tools for developing metacognitive awareness in L2 learners, noting their growing relevance for younger and less-experienced language users.

2.4. Developmental and Cultural Considerations

While think-aloud strategies have been extensively studied with adult learners, their application with children aged 6 - 8 raises specific developmental considerations. Larkin (2010) identifies metacognitive development in young children as highly sensitive to instructional context: children receiving explicit, repeated exposure to metacognitive strategy modelling show greater evidence of self-regulatory behaviour than those in transmission-oriented classrooms. Whitebread et al. (2009) further note that peer interaction and collaborative tasks create naturalistic conditions for metacognitive verbalisation. Asking young learners to simultaneously produce language and monitor their thinking may create cognitive overload if not carefully mediated (Dent & Koenka, 2016), suggesting that staged scaffolding which separates comprehension from production is essential for this age group.

Cultural factors further shape learner engagement in the Chinese context. Hu (2002) analyses the tension between communicative language teaching and the examination-oriented, teacher-authoritative tradition prevalent in Chinese schools, arguing that accuracy-focused pedagogy may generate affective barriers for learners asked to verbalise incomplete thinking. Bao (2014) similarly identifies face-saving and deference to the teacher as sources of reticence, while Dan and Bai (2025) found, in a study of 436 Chinese primary-age EFL learners, that positive classroom relationships with teachers and peers were associated with greater use of self-regulated and metacognitive strategies such as monitoring, an association partly explained by the more positive emotions those relationships fostered.

3. Research Design

3.1. Research Design and Philosophical Approach

This study adopts a qualitative practitioner enquiry approach, which Cochran-Smith and Lytle (2009) describe as systematic, intentional study by teachers of their own classroom practice. Practitioner enquiry foregrounds the teacher-researcher’s professional context as the site of knowledge generation, enabling critical reflection on how pedagogical decisions shape learning outcomes (McNiff, 2017). The enquiry does not seek statistical generalisation but rather context-specific understanding of how the think-aloud strategy operates with these particular learners in this particular setting. A case study approach is appropriate precisely because I am not trying to make a claim about all ESL learners. I am trying to understand what happened with these three learners, in this classroom, in this learning centre. Yin (2018) argues that case study is the right design when the question is about how or why something happens, and when the phenomenon cannot be meaningfully separated from its context. That is exactly the situation here: the learners’ think-aloud behaviour cannot be separated from the specific tasks, the relationship between teacher and learners, or the cultural setting in which the sessions took place. The case is the implementation of the think-aloud strategy across two consecutive teaching sessions, with three purposively sampled learners constituting embedded sub-cases.

Two consecutive sessions were judged sufficient for the scope of this enquiry, which asks whether the think-aloud strategy can be observed to operate at all in this setting, rather than whether its effects are durable over time; a single paired comparison is adequate to establish the former, even though it cannot establish the latter. This scope inevitably limits what the change in scores between Session 1 and Session 2 can be taken to show. Two alternative explanations cannot be ruled out from a two-session design. First, a novelty effect is possible: learners may have engaged more readily with think-aloud in Session 2 simply because the activity, the researcher’s attention, or the departure from routine lesson format was still new, rather than because metacognitive capacity had genuinely developed. Second, a practice effect is possible: gains in domain scores may partly reflect increasing familiarity with the specific vocabulary set, task format, and rating procedure used across the two sessions, rather than a transferable change in self-regulation. The enquiry cannot statistically separate these explanations from a genuine strategy effect within a two-session window, and the improvements reported in Section 4 are accordingly interpreted as indicative of early-stage, promising change rather than as evidence of consolidated or durable metacognitive development. This qualification is revisited in the concluding limitations in Section 5.

3.2. Participants and Context

Three Chinese ESL learners aged between 6 and 8, attending a private English learning centre in China, participated in this study. Participants were purposively selected (Merriam & Tisdell, 2016) to represent three proficiency bands: higher (L1), mid (L2), and lower (L3), enabling examination of differential strategy uptake across ability levels. Proficiency band was assigned by the teacher-researcher on the basis of the learners’ performance across the preceding four weeks of regular centre lessons, drawing on routine in-centre assessment records (weekly vocabulary and grammar checks) together with the teacher-researcher’s own ongoing classroom judgement of each learner’s spoken and written English; no external or standardised proficiency test was used. This basis for banding is a limitation of the sampling rationale, since teacher judgement of this kind is not independently verified, and is noted as such in Section 4. All participants are referred to exclusively by pseudonym, and the name and location of the learning centre have been redacted from all research documentation. The teaching sessions took place within the learners’ regular English lessons covering Unit 4 of the standard curriculum: house vocabulary, prepositions of place, and subject-pronoun agreement. Session 1 (18 May 2026) addressed house vocabulary and prepositions of place, and Session 2 (19 May 2026) extended to kitchen vocabulary and subject-pronoun agreement (it/they). The think-aloud strategy was integrated into routine instructional activities rather than implemented as an isolated intervention, supporting the ecological validity of the findings. (Table 1)

Table 1. Participant profile and structured research log totals across sessions (full domain scores and field notes are presented in Appendix C).

3.3. Data Collection Methods

Two complementary methods were employed across both sessions. Cohen et al. (2018) describe methodological triangulation as using more than one source of data to strengthen confidence in findings: when both the structured research log and the child questionnaire point toward the same conclusion about a learner, that conclusion carries more evidential weight than either source would alone.

The primary instrument was a structured research log developed specifically for this enquiry (the blank instrument is reproduced in Appendix A). As the teacher-researcher, observation notes were completed immediately following each session, capturing behavioural ratings across four domains: Verbal Metacognitive Engagement (V1 to V5); Self-Monitoring and Self-Regulation (M1 to M5); Task Engagement and Strategy Use (E1 to E4); and Affective and Social Indicators (A1 to A3). Each item was rated on a three-point scale (0 = not observed; 1 = attempted; 2 = clearly evident), with qualitative field notes documenting specific learner verbalisations, moments of self-correction, and instances of hesitation or reluctance. Structured observation is appropriate for research involving young learners, for whom self-report instruments present reliability challenges due to limited reflective language skills (Whitebread et al., 2009). To mitigate the observer bias inherent in the teacher-researcher role (Cohen et al., 2018), behavioural indicators were operationally defined in advance, notes were completed post-session, and a reflexive positionality note was recorded for each session.

The secondary instrument was the Child-Adapted Learner Questionnaire, administered at the close of each session (the blank instrument is reproduced in Appendix B). The instrument used a three-point smiley-face Likert scale and six bilingual (English and Mandarin) items organised across two sections: Section 1 (Q1 to Q3) addressed learner experience of thinking aloud, covering enjoyment, perceived helpfulness for task understanding, and ease of speaking in English; Section 2 (Q4 to Q6) addressed confidence, perceived usefulness for error-finding, and willingness to repeat the strategy. Visual Likert tools have been validated as appropriate for eliciting attitudinal data from young learners (Muijselaar et al., 2021). Responses are summarised in Table 2.

Table 2. Child-adapted learner questionnaire responses (S1 = Session 1; S2 = Session 2) (full completed responses are presented in Appendix D).

3.4. Data Analysis

Data were analysed using reflexive thematic analysis as developed by Braun and Clarke (2022). Unlike frameworks that apply pre-existing codes, this approach starts from the data: patterns are identified, named, grouped into themes, and checked back against the data. Braun and Clarke argue that the researcher’s perspective is not a problem to be controlled. It is an active part of how meaning is made. This is particularly appropriate for practitioner enquiry, where insider knowledge of the learners and context is an analytical resource rather than a source of contamination (McNiff, 2017). Analysis proceeded through familiarisation, initial coding, theme construction, review, definition, and interpretive write-up contextualised within the literature. Questionnaire responses were coded numerically (Yes = 2, A little = 1, No = 0) to enable cross-session and cross-learner comparison alongside observation data.

Because the research log domains and questionnaire items were pre-structured rather than open-ended, thematic analysis here worked across two linked data forms rather than from unstructured transcript alone: the numerical domain and item scores indicated where a pattern existed and how strong it was, while the qualitative field notes and raw verbalisations indicated what that pattern actually looked like in practice. A worked example illustrates the process. The raw field note for L3 in Session 2 records the utterance “It behind the… It’s… It’s behind the armchair,” alongside a Self-Monitoring and Self-Regulation (M1 to M5) score that had risen from 5/10 to 6/10. This note was initially coded descriptively as “mid-utterance self-correction with teacher-independent repair.” Comparable codes recurred across L1’s Session 2 note (“Wait, lamp is one thing, so it’s ‘It’s,’ not ‘They’re.’ Let me fix that”) and L2’s Session 2 note on the missed Read and Write item, where self-correction followed a prompt rather than preceding it. Grouping these codes by whether self-correction was unprompted, partially prompted, or absent, and cross-checking the grouping against the Domain 2 score pattern across all six logs, produced Theme 2 (Self-Monitoring and Self-Regulatory Behaviours Developed Differentially). The same logic of coding first from field notes, then checking the resulting groupings against the pre-structured domain and questionnaire scores for confirmation or disconfirmation, was applied consistently to generate all five themes, rather than themes being read directly off the numerical scores.

3.5. Ethical Considerations

Ethical approval was obtained through the University of Cumbria ethics review process prior to data collection. Written informed consent was obtained from parents and guardians, and verbal assent was sought from the learners themselves. Participation was entirely voluntary, and parents were informed that withdrawal was possible at any point within seven days of participation, without consequence for their child’s learning or relationship with the teacher. All identifying data have been redacted from research records, including participants’ names, ages, gender, and the name and location of the learning centre. This anonymisation applies specifically at participant level: the three child learners are identified only by pseudonym throughout, and no detail permitting their individual identification is disclosed. It does not extend to the institutional and supervisory level, where identification was retained deliberately and with consent: the university supervisor, the learning centre’s headmaster, and the learning centre’s owner are named in the Acknowledgements because their permission and support for the enquiry are a matter of institutional record and because none of these individuals is a research participant whose identity requires protection. The learning centre itself remains unnamed and its location redacted throughout the main text and appendices, consistent with the terms of the ethics approval. All electronic data are stored securely on the researcher’s University of Cumbria OneDrive account and will be permanently deleted no later than 30 September 2026.

4. Findings and Discussion

Analysis of the structured research log data and questionnaire responses across both sessions produced five themes. Each theme is presented below with the observation data as the primary evidence and the questionnaire data as a secondary triangulating source, providing the learner’s own perspective on the strategy alongside the researcher’s behavioural observations. The full domain score summaries and field notes underlying this analysis are presented in Appendix C, and the full completed questionnaire responses in Appendix D; excerpts are quoted below where they illustrate a theme directly.

Theme 1: The Think-Aloud Strategy Promoted Emerging Metacognitive Awareness

The most consistent finding was that the think-aloud strategy created observable opportunities for learners to externalise and begin to develop metacognitive awareness. This was most clearly evident in L1, who demonstrated unprompted, sustained verbal monitoring from Session 1. A notable example arose in Session 2, when L1 self-corrected aloud: Wait, lamp is one thing, so it’s ‘It’s,’ not ‘They’re.’ Let me fix that (see Appendix C2). This utterance reflects active monitoring of grammatical accuracy in real time, consistent with Flavell’s (Flavell, 1979) model of metacognitive monitoring and Ericsson and Simon’s (Ericsson & Simon, 1993) account of concurrent verbal protocols as evidence of cognitive regulation. For L2 and L3, verbalisation was less spontaneous in Session 1, requiring teacher prompting to sustain. By Session 2, however, both learners showed measurable growth: L2 narrated independently during the kitchen vocabulary task, and L3 began offering partial procedural descriptions during the Read and Match activity (see Appendix C4 and Appendix C6 respectively). These shifts support the view that metacognitive awareness is not a binary capacity but a developmental trajectory that the think-aloud strategy can meaningfully support with sufficient scaffolding (Larkin, 2010).

All three learners reported in Session 2 that thinking aloud had helped them understand the task (Q2: Yes for all), representing a clear shift from Session 1 in which L3 responded only ‘A little.’ This convergence between observation data and learner perception strengthens the finding that think-aloud instruction promoted emerging metacognitive awareness across the group.

Theme 2: Self-Monitoring and Self-Regulatory Behaviours Developed Differentially

The data reveal that the think-aloud strategy facilitated self-monitoring with significant variation across proficiency levels. L1 demonstrated strong and consistent self-monitoring in both sessions, scoring 8/10 and 9/10 on M1 to M5 respectively, with self-correction observed multiple times unprompted (see Appendix C1 and Appendix C2). L1’s behaviour reflects what Zimmerman (2002) describes as characteristic of a proficient self-regulated learner: she noticed her own error without prompting, revised it, and continued through the monitoring-and-adjustment cycle fluidly. L2 showed a more transitional pattern: field notes from Session 1 record L2 saying the fireplace is in front of? No, next to the armchair, indicating that think-aloud was surfacing a self-corrective process that might otherwise have remained invisible (see Appendix C3). By Session 2, L2’s self-monitoring score rose from 7/10 to 9/10, and self-correction was now observed unprompted across multiple items (see Appendix C4). L3 presented the most complex trajectory: spontaneous self-correction was not observed in Session 1, though L3 consistently sought clarification (M5 = 2 in both sessions), reflecting a form of meta-awareness that understanding was incomplete. By Session 2, partial self-correction was observed when L3 revised ‘It behind the’ to ‘It’s behind the armchair,’ (see Appendix C6) consistent with Teng’s (Teng, 2020) finding that metacognitive regulation develops incrementally and is particularly sensitive to instructional scaffolding.

The questionnaire provides triangulating evidence for this theme (see Appendix D). On Q5, both L1 and L2 responded ‘Yes’ in both sessions, reflecting consistent perception of the strategy’s self-corrective value. L3 shifted from ‘A little’ in Session 1 to ‘Yes’ in Session 2, aligning with the emergence of self-corrective behaviour in the observation data and strengthening the finding that the strategy meaningfully supported self-monitoring across all proficiency levels.

Theme 3: Learner Proficiency Shaped Strategy Use

Proficiency level was the most consistent differentiating variable in strategy uptake. L1 achieved near-perfect scores in Verbal Metacognitive Engagement (9/10 and 10/10) and extended think-aloud into peer interaction, independently generalising the it/they pronoun distinction to novel contexts (see Appendix C1 and Appendix C2). L2 showed the most visible qualitative shift across sessions: every item in Session 2 for which L2 narrated the reasoning produced a correct response, while the single silently completed item was missed, directly supporting the claim that verbalisation functions as a live comprehension-monitoring tool. For L3, limited English production created a dual constraint, requiring simultaneous language production and metacognitive monitoring. Despite this, L3’s verbal metacognitive engagement score increased from 3/10 in Session 1 to 8/10 in Session 2, the largest single-domain gain across the study (see Appendix C5 and Appendix C6). L3’s visual support use (E2 = 2 in both sessions) demonstrated strong comprehension not yet fully translating into verbal output, suggesting lower-proficiency learners benefit from staged scaffolding that separates comprehension from production before requiring concurrent think-aloud narration. On Q3, L1 and L2 progressed from ‘A little’ to ‘Yes’ on ease of speaking, while L3 remained at ‘A little’ in Session 2, confirming that the linguistic barrier to verbalisation persisted.

Theme 4: Affective and Cultural Factors Shaped Participation

A consistent pattern was the role of affective and culturally shaped factors in mediating willingness to engage. L3 entered Session 1 with closed body language for approximately the first thirty minutes, despite demonstrating comprehension through correct picture-pointing (see Appendix C5). This suggests that the affective barrier to verbalisation operated independently of comprehension capacity. The observation is consistent with Hu’s (Hu, 2002) analysis of tension between communicative pedagogy and accuracy-focused culture in Chinese educational settings, and with Bao’s (Bao, 2014) identification of face-saving as a source of reticence. Dan and Bai (2025) extend this to young Chinese EFL learners specifically, finding that the quality of learners’ relationships with teachers and peers shapes their strategy use partly through the emotions those relationships generate. The questionnaire data provide a vivid window into L3’s affective experience (see Appendix D): in Session 1, L3 responded ‘No’ to Q1 (enjoyment), ‘No’ to Q4 (confidence), and ‘A little’ to Q6 (willingness to repeat). By Session 2, L3 responded ‘Yes’ to Q1 and Q6, and ‘A little’ to Q4, indicating a meaningful reduction in affective resistance over a single session. L2 similarly shifted from ‘A little’ to ‘Yes’ on Q4, corroborated by observation data showing A2 (hesitancy) rated as not observed in Session 2. Vandergrift and Goh (2012) note that metacognitive strategy use is mediated by confidence and willingness to take communicative risks, a position strongly supported by the evidence gathered here.

Theme 5: Teacher Scaffolding Was an Indispensable Condition for Strategy Access

The centrality of teacher scaffolding to strategy effectiveness was the most significant finding for pedagogical practice. Without scaffolding, think-aloud did not occur spontaneously for L3, and occurred only partially for L2. Scaffolding took several forms: explicit teacher modelling of think-aloud at the start of each session using utterances such as ‘I think this means…’ and ‘I am not sure, so I will look again’; sentence-starter prompt cards; guided questioning during independent work; and consistent acknowledgement of verbalisation attempts regardless of accuracy. These techniques align with Vygotsky’s (Vygotsky, 1978) ZPD framework, in which adult mediation provides temporary support through which learners access competencies beyond their current independent capacity. Following the introduction of sentence-starter prompts at the opening of Session 2, L3 began verbalising within the first five minutes, compared to approximately thirty minutes in Session 1 (see Appendix C5 and Appendix C6). By Session 2, L3 also began using self-directed repetition during the partner game, repeating ‘They’re on the rug’ twice in succession, which field notes interpret as deliberate rehearsal and an early indicator of emerging independent metacognitive strategy use. In Session 2, all three learners responded ‘Yes’ to Q6 (willingness to repeat the activity), compared to Session 1 in which L3 responded ‘A little’ (see Appendix D). This unanimity suggests that scaffolded delivery had, within two sessions, created a positive learner orientation towards think-aloud practice. Fisher and Frey (2018) describe this progression as a gradual release of responsibility, moving from full teacher demonstration through guided joint practice to independent learner use. By Session 2, the learners had begun to move from the first stage toward the second.

A methodological caveat follows directly from this theme. Think-aloud was never delivered as an isolated variable in this enquiry: it was always accompanied by teacher modelling, sentence-starter prompt cards, guided questioning, and consistent acknowledgement of attempts. Because these elements were introduced together as a single scaffolded package, particularly for L3 from the start of Session 2, the gains reported across Themes 1 to 4 cannot be attributed to verbalisation in isolation. It is equally plausible that the observed improvement in metacognitive engagement, self-monitoring, and affective participation reflects the scaffolding package as a whole, with think-aloud as one component among several, rather than the act of verbalisation carrying the effect on its own. This enquiry’s design, using a single integrated intervention across two sessions, cannot separate these components empirically; doing so would require withholding one element of scaffolding while retaining the others, which was neither ethically nor practically appropriate in this classroom setting. The claims made in this study should therefore be read as claims about the effectiveness of a scaffolded think-aloud package, not about think-aloud as a standalone technique.

It is important to acknowledge what the findings collectively do not show. Within two sessions, none of the three learners demonstrated fully spontaneous, sustained metacognitive narration across all task types without prompting. Dent and Koenka (2016) note that measurable self-regulatory gains typically require multiple exposures over a sustained period, suggesting the evidence reflects promising early-stage development rather than consolidated metacognitive competence. The reliability of the research log data is also constrained by the teacher-researcher dual role: despite pre-defined behavioural indicators and post-session recording, the possibility of confirmation bias cannot be entirely eliminated. These limitations do not invalidate the findings but qualify the confidence with which conclusions can be drawn.

5. Conclusion

This practitioner enquiry examined whether, and in what ways, the think-aloud strategy facilitates metacognitive awareness and high-quality teaching and learning among young Chinese ESL learners in a private learning centre. The findings, drawn from structured research log data and child-adapted questionnaire responses across two sessions with three purposively sampled learners, indicate that the strategy holds meaningful pedagogical potential, though its effectiveness is neither uniform nor automatic.

L1 demonstrated the highest and most autonomous engagement, producing spontaneous verbal monitoring, self-correction, and cross-linguistic metacognitive reasoning. L2 showed the most visible qualitative shift between sessions, with a direct correlation between verbalisation and task accuracy, and questionnaire confidence shifting from “A little” to “Yes.” L3 made the most proportional progress, with verbal metacognitive engagement rising from 3/10 to 8/10, and questionnaire responses shifting from predominantly negative in Session 1 to predominantly positive in Session 2.

Three pedagogical implications emerge. First, think-aloud instruction must be explicitly modelled, consistently scaffolded, and progressively released, particularly for lower-proficiency learners. Second, the affective and cultural climate of the classroom is as important as the strategy itself: the shift in L3’s questionnaire responses across a single session demonstrates that even brief positive experiences can begin to reduce affective resistance. Third, peer-modelling through structured partner activities creates naturalistic conditions for metacognitive verbalisation that should be deliberately incorporated into lesson design.

It should be emphasised that these implications concern a scaffolded think-aloud package, comprising teacher modelling, prompt cards, guided questioning, and verbalisation together, rather than verbalisation in isolation. This enquiry cannot determine what proportion of the observed gains is attributable to think-aloud specifically as opposed to the surrounding scaffolding, and this distinction should be tested directly in future research designs that vary the components independently.

This enquiry is limited by its small sample size and two-session timeframe. Future research should extend the intervention across a longer period and incorporate audio recording for greater precision. Nonetheless, the alignment between observation and questionnaire data provides encouraging evidence that structured think-aloud instruction, when carefully designed and responsively delivered, represents a viable and valuable metacognitive approach in early ESL settings.

Acknowledgements

The author wishes to thank Pippa Leslie (University of Cumbria) for her supervision and scholarly guidance throughout this study. Sincere thanks are also extended to Headmaster Weijie Gu for granting institutional access and supporting the research, and to Huang PeiJun for permitting data collection within his learning centre. The author is grateful to LinLin Zhou for her contextual expertise on Chinese educational practices, which informed the framing of this study. Finally, the author acknowledges the young learners whose participation made this research possible.

Appendices

The following appendices contain supporting materials for this practitioner enquiry. They are provided to evidence the data collection instruments designed for this study and to present the raw data on which the findings are based.

Appendix A: Structured Research Log (Blank Instrument)

This appendix presents the blank Structured Research Log as designed by the researcher and used as the primary data collection instrument across both sessions. The log was completed by the teacher-researcher immediately following each session in relation to each of the three participant learners.

PART A—SESSION INFORMATION

PART B—RATING SCALE KEY

PART C—BEHAVIOURAL OBSERVATION DOMAINS

Domain 1: Verbal Metacognitive Engagement (V1 to V5)—spontaneous verbalisation; English narration; step-by-step description; expression of uncertainty; mirroring of teacher modelling. Domain 2: Self-Monitoring and Self-Regulation (M1 to M5)—task revisiting; unprompted self-correction; response checking; error-responsive strategy change; clarification requests. Domain 3: Task Engagement and Strategy Use (E1 to E4)—independent task attempt; verbalised use of visual support; sustained on-task focus; task completion awareness. Domain 4: Affective and Social Indicators (A1 to A3)—comfort and willingness to verbalise; hesitancy or reluctance; peer activity engagement. Each indicator rated 0, 1, or 2.

PART D—QUALITATIVE FIELD NOTES

Fields: 1) Notable examples of think-aloud use (direct quotes where possible). 2) Moments where the learner struggled with the strategy and what happened. 3) Observed impact on task performance or comprehension. 4) Researcher positionality note. 5) Any other relevant observations.

PART E—DOMAIN SCORE SUMMARY

Maximum scores: Domain 1 = 10, Domain 2 = 10, Domain 3 = 8, Domain 4 = 6. Total maximum = 34. Scores summed per domain and compared across sessions.

Appendix B: Child-Adapted Learner Questionnaire (Blank Instrument)

This appendix presents the blank Child-Adapted Learner Questionnaire as designed by the researcher and administered to all three learners at the close of each session. The instrument is bilingual (English and Mandarin Chinese) and uses a three-point smiley-face Likert scale: Yes (happy face), A little (neutral face), No (sad face).

Section 1: Talking Out Loud

Q1. I liked saying my thinking out loud during English class. / 我喜欢在英语课上大声说出我的想法。

Q2. Saying my thinking out loud helped me understand the English task. / 大声说出我的想法帮助我理解了英语任务。

Q3. It was easy for me to talk out loud in English. / 对我来说，用英语大声说话很容易。

Section 2: My Learning

Q4. I felt confident when I talked out loud about the English task. / 当我大声谈论英语任务时，我感到很自信。

Q5. Talking out loud helped me find my mistakes. / 大声说出想法帮助我发现了自己的错误。

Q6. I want to use this talking-out-loud activity again in English class. / 我想在英语课上再次使用大声说话的活动。

Each question answered by ticking one box: Yes / A little / No. The questionnaire takes approximately 3 to 6 minutes to complete.

Appendix C: Completed Structured Research Logs

This appendix presents the domain score summaries and selected field notes from all six completed structured research logs (three learners across two sessions each). Full qualitative field notes are summarised under each learner. Raw domain scores are presented in tabular form to support the data presented in Section 3.2 and the findings discussed in Section 4.

C1: L1 (Higher Proficiency)—Session 1|18 May 2026

Activity: Unit 4 House Vocabulary (Listen, Point and Say) and Grammar 1: Prepositions of Place (True/False and Look and Write). Duration: 60 minutes.

C2: L1 (Higher Proficiency)—Session 2|19 May 2026

Activity: Unit 4 Vocabulary 2: Kitchen Items and Grammar 2: Subject-Pronoun Agreement it/they. Duration: 60 minutes.

C3: L2 (Mid Proficiency)—Session 1|18 May 2026

Activity: Unit 4 House Vocabulary (Listen, Point and Say) and Grammar 1: Prepositions of Place (True/False and Look and Write). Duration: 60 minutes.

C4: L2 (Mid Proficiency)—Session 2|19 May 2026

Activity: Unit 4 Vocabulary 2: Kitchen Items and Grammar 2: Subject-Pronoun Agreement it/they. Duration: 60 minutes.

C5: L3 (Lower Proficiency)—Session 1|18 May 2026

Activity: Unit 4 House Vocabulary (Listen, Point and Say) and Grammar 1: Prepositions of Place (True/False and Look and Write). Duration: 60 minutes.

C6: L3 (Lower Proficiency)—Session 2|19 May 2026

Activity: Unit 4 Vocabulary 2: Kitchen Items and Grammar 2: Subject-Pronoun Agreement it/they. Duration: 60 minutes.

Appendix D: Completed Child-Adapted Learner Questionnaire Responses

This appendix presents the completed questionnaire responses for all three learners across both sessions, as read directly from the handwritten questionnaire forms. Responses are colour-coded: green = Yes, amber = A little, red = No.

Table D1. Completed questionnaire responses (S1 = Session 1; S2 = Session 2; Green = Yes; Amber = A little; Red = No).

Questionnaire dates: L1 Session 1 = 18 May 2026; L1 Session 2 = 19 May 2026; L2 Session 1 = 18 May 2026; L2 Session 2 = 19 May 2026; L3 Session 1 = 18 May 2026; L3 Session 2 = 19 May 2026. All questionnaires were administered by the teacher-researcher at the close of each lesson. Responses were self-reported by each learner independently using a smiley-face tick-box format.

Note on L3 questionnaire responses: In Session 1, L3 responded No to Q1 (enjoyment), A little to Q2 (helpfulness), No to Q3 (ease of speaking), No to Q4 (confidence), A little to Q5 (error-finding), and A little to Q6 (willingness to repeat). By Session 2, L3 responded Yes to Q1, Q2, and Q6, and A little to Q3, Q4, and Q5. This shift is consistent with the observation data presented in Appendix C and the findings discussed in Section 4, Theme 4.

Conflicts of Interest

The authors declare no conflicts of interest regarding the publication of this paper.

References