Limitations of the Study

Although the study provided useful insights into how a repeated set of speaking activities can influence young learners' willingness to communicate, several limitations must be acknowledged. These limitations affect how the findings should be interpreted and to what extent they can be applied to other classrooms. Reporting them openly is part of the trustworthiness of small-scale classroom action research (Cohen, Manion & Morrison, 2018, pp. 245–250).

The study involved only four learners in one Grade 2 EFL class in a Polish primary school. This small number means that the findings cannot be generalised to other classes, age groups, or educational contexts. The four learners were not selected to represent a larger population; they were the actual members of the class taught by the researcher.

However, this is consistent with the nature of action research, which prioritises depth over breadth and aims to produce transferable, not statistically representative, findings (Cohen et al., 2018, p. 442). Other teachers working with similar age groups in similar low-exposure EFL contexts may find that the observed patterns resonate with their own classrooms.

The researcher was also the classroom teacher. This dual role made the intervention possible because it allowed close access to learners, but it also creates a potential source of observer bias: the same person designed the activities, delivered them, and recorded learner behaviour. Learners may also have responded differently because their teacher was present.

To reduce this risk, the study used triangulation across four data sources: structured observation with pre-defined indicators, child self-report questionnaires (SPCC and FLCA), short activity feedback forms, and teacher field notes. The observation sheet used countable behavioural indicators rather than subjective judgements, and the questionnaires captured the learners' own voices independently of the teacher's interpretation.

The intervention lasted only eight weeks, organised into three action research cycles. This is too short to capture deep, long-term change in learners' WTC or in their self-concept as English speakers. Some observed improvements — such as more frequent question-asking or longer answers — may reflect short-term familiarity with the activity set rather than stable shifts in willingness to communicate.

Eight weeks also makes it difficult to fully separate the effect of the intervention from natural developmental growth in young children. Children at ages 7–8 are at a stage where speaking skills and social confidence develop quickly. A longer study, ideally across a full school year, would help distinguish intervention effects from maturation.

The study did not include a control or comparison group. Every learner in the class received the intervention, so it is not possible to compare their development with a parallel class that did not receive it. This limits any causal claim about the effect of the speaking activity set on WTC.

However, a control group was neither feasible nor ethically appropriate in this context. The class had only four learners, and withholding planned speaking activities from comparable learners would have been pedagogically and ethically problematic. Action research deliberately accepts this trade-off in exchange for an authentic, integrated teaching context (Cohen et al., 2018, pp. 447–449).

Both the SPCC and FLCA questionnaires were originally designed for older learners and adults. For this study they were adapted into a child-friendly form: items were shortened, wording simplified, emoji-face Likert points used in place of numerical scales, and items read aloud by the teacher.

While these adaptations were necessary for 7–8 year-olds, they reduce the direct comparability between this study's SPCC/FLCA scores and those reported in the wider literature. The scores should therefore be read as relative indicators of pre–post change within the group, not as absolute measurements of communicative competence or anxiety.

The intervention was carried out in a very specific setting: a Polish primary school using Learning Lands 2, a fixed small group of four learners, a one-hour weekly English lesson, and a Vietnamese teacher-researcher working in a Polish-speaking context. The dynamics of this setting cannot easily be reproduced elsewhere.

Cultural factors — including learners' first language background, school routines, and broader attitudes toward speaking English — would differ in other contexts and could change which speaking activities work best. The findings should therefore be read as context-bound evidence about WTC dynamics, useful as a reference point for similar settings rather than as universal prescription.

The study measured observable WTC-related behaviour — speaking first, asking questions, producing longer answers, and talking to peers — rather than internal willingness as a psychological state. A learner may feel highly willing to speak but stay silent because of a dominant peer, an unclear task, or a momentary distraction. The opposite is also possible: learners may speak because they were prompted by classroom routines rather than because they truly wanted to.

The four behavioural indicators therefore capture the output of the WTC pyramid (Layer I — actual L2 use), but only indirectly the inner layers. The SPCC and FLCA questionnaires partially compensate for this by accessing learners' self-perceptions, but no instrument can fully access the moment-by-moment psychological state described by MacIntyre et al. (1998).

These limitations point toward several directions for future research: