Reading aloud has long been discussed as a pedagogical activity in second and foreign language classrooms. Earlier discussions emphasized its usefulness for pronunciation practice, oral fluency development, and learners’ awareness of phonological forms. Although RA has sometimes been criticized as mechanical or insufficiently communicative, later work has argued that it can become meaningful when embedded in broader communicative or task-based instruction.

From this perspective, RA offers learners a relatively structured form of oral production. Because the linguistic content is already available in written form, learners can focus more directly on articulation, pacing, intonation, and delivery. This makes RA particularly relevant for learners with limited speaking confidence or limited experience producing English orally. In classroom-based EFL settings, RA has also been used as a bridge between individual practice and more public forms of oral performance.

Studies conducted in university EFL contexts suggest that RA can support learner engagement and perceived oral development when it is used consistently over time. In the Korean context, RA-based instruction using authentic materials such as English TV news or translation-supported tasks has also been reported to help learners become more comfortable with oral delivery (Chong, 2019, 2021).

Task repetition provides an important theoretical explanation for the value of repeated RA. Bygate (2001) argued that repeating the same task allows learners to shift their attention from content to controlling language during performance. In oral reading contexts, repeated exposure to the same text may reduce uncertainty, stabilize delivery, and support fluency.

Empirical studies support this interpretation. Gorsuch and Taguchi (2010) found that repeated reading contributed to gains in fluency and learner confidence in EFL contexts. More recent studies have also shown that digitally supported RA protocols can improve oral fluency, suggesting that repetition combined with structured practice environments may strengthen oral performance (Rodríguez-Fuentes et al., 2025).

These studies support the idea that repeated RA may help reduce at least some forms of affective burden by increasing familiarity with task demands. At the same time, fluency gains do not automatically imply global anxiety reduction. The affective consequences of repeated oral reading may vary depending on task framing and classroom context.

Writing is often used as a pre-speaking activity because it allows learners to organize ideas and language before speaking. In Korean university EFL contexts, writing-based preparation has also been shown to support learners’ readiness for subsequent oral performance, although its affective consequences may vary depending on how writing is integrated into the task sequence (Chong, 2022). This suggests that writing can function as a cognitive scaffold while also shaping learners’ awareness of evaluation and accuracy.

However, writing may also intensify self-monitoring. Research on writing anxiety indicates that learners may become overly attentive to grammatical accuracy, wording, and evaluation while composing written text (Kara, 2013). This concern remains relevant in university EFL settings, where recent research suggests that speaking- and writing-related anxiety may interact rather than operate as fully separate affective experiences, especially in tasks that combine written preparation with oral performance (Özdemir & Seçkin, 2025). When oral performance follows immediately after writing, the benefits of planning may coexist with increased pressure related to correctness and self-presentation.

This issue appears particularly relevant in RA-based instruction that incorporates writing, reading aloud, and recording. A previous classroom-based study found that writing supported preparation but also seemed to heighten sensitivity to assessment and accuracy in subsequent oral performance (Chong, 2022). Taken together, these findings suggest that writing may operate as a double-edged preparatory stage: it can increase readiness, but it may also increase evaluative awareness.

Foreign language classroom anxiety has been widely examined as a central affective factor in language learning. Horwitz et al. (1986) defined FLCA as a situation-specific construct that includes communication apprehension, test anxiety, and fear of negative evaluation. Subsequent research showed that anxiety can interfere with cognitive processing during language tasks, particularly when learners must produce language in real-time (MacIntyre & Gardner, 1994).

Speaking-related tasks are often experienced as especially anxiety-provoking. Matsuda and Gobel (2004) found that anxiety predicted performance in foreign language classrooms, and more recent work has suggested that task conditions and learner psychology can interact in complex ways during oral performance (Jin et al., 2021; Mora et al., 2024).

For the present study, the key point is that FLCA is multidimensional. An instructional sequence that supports communication-related comfort may not necessarily reduce evaluation-related anxiety. Recent review research likewise suggests that foreign language anxiety research is increasingly moving toward more skill-specific, context-sensitive, and pedagogically situated analyses rather than treating anxiety as a single undifferentiated construct (Yu, 2024). This distinction is especially important when comparing RA-only instruction with instruction that integrates writing before RA.

Taken together, existing research suggests that RA and task repetition can support oral development and may reduce some aspects of performance burden, while pre-task writing may function differently depending on task design and evaluative framing. However, several gaps remain. First, few classroom-based studies have directly compared RA-only instruction with writing-integrated RA in terms of changes in FLCA over time. Second, prior studies have often emphasized total anxiety scores rather than examining whether subcomponents of FLCA respond differently to instructional sequencing.

Building on earlier RA-focused classroom research and established models of FLCA, the present study addresses these gaps by comparing RA and W+RA instructional conditions using a pre-post design. By examining both total FLCA and its subdimensions, this study aims to provide a more differentiated account of how instructional task sequencing relates to learners’ affective experiences in EFL oral instruction. More specifically, the study contributes a classroom-based, subscale-sensitive comparison of two semester-long instructional configurations in university general English education rather than proposing a new stand-alone model of foreign language anxiety. This approach also aligns with recent work re-examining the interpretation and application of the FLCAS in present-day EFL contexts (Khreisat, 2024).

This study adopted a quasi-experimental, non-randomized pre-post design to examine how two oral task-sequencing models were associated with changes in learners’ FLCA over time. The study was conducted in the context of a 15-week university elective English course. As students enrolled through the university’s standard registration system, random assignment was not feasible. The study therefore relied on intact classes. Because the study relied on self-selected intact classes rather than random assignment, pre-existing differences between groups cannot be ruled out. Accordingly, the findings should be interpreted as associative rather than strictly causal.

Two instructional conditions were examined: reading aloud (RA) and writing-then-reading-aloud (W+RA). To address the research questions, FLCA was measured with the full 33-item FLCAS. Changes in anxiety were examined both within instructional conditions across time and between conditions in terms of change patterns. In addition, subscale-level analyses were conducted to identify which dimensions of FLCA - CA, TA, and FNE - were most sensitive to task sequencing effects.

Because the two instructional conditions differed not only in preparatory sequence but also in task materials and discourse orientation, the findings should be interpreted as comparisons between two naturally occurring instructional configurations rather than as evidence of the isolated effect of writing alone. The RA condition was based primarily on English TV news texts, whereas the W+RA condition involved practical speaking topics associated with OPIc- and TOEIC-speaking-style performance. Accordingly, the observed differences should be understood as reflecting multiple features of the instructional configurations rather than task sequence alone.

The participants were undergraduate students enrolled in two elective English courses at a private university in South Korea during the spring semester of 2022. A total of 114 students completed both the pre- and post-surveys and were included in the final analysis: 59 in the RA group and 55 in the W+RA group.

Students self-selected into courses through the university’s standard enrollment system, resulting in self-selected class membership rather than random assignment. The participants represented a range of academic majors and years of study, reflecting the typical composition of elective English courses at the institution. All participants had previously completed the university’s required freshman English courses. Based on their placement information, most students were broadly within the intermediate range typical of non-English majors at the university.

As shown in Table 1, the present study used the FLCAS total scale together with three conceptually relevant subdimensions for analysis. FLCA was measured using the Foreign Language Classroom Anxiety Scale (FLCAS; Horwitz et al., 1986). The instrument consists of 33 items rated on a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree), with higher scores indicating higher levels of anxiety. Negatively worded items were reverse-coded prior to analysis.

In addition to the total FLCAS score, this study examined three commonly discussed dimensions derived from Horwitz et al.’s conceptualization of FLCA: Communication Apprehension (CA; 8 items), Test Anxiety (TA; 6 items), and Fear of Negative Evaluation (FNE; 7 items). In the present study, these subdimensions were treated as theory-informed analytical dimensions based on Horwitz et al.’s conceptualization of FLCA rather than as newly validated independent constructs. These subscale analyses were conducted to examine whether different dimensions of anxiety responded differently to instructional sequencing.

The FLCAS was administered twice: once near the beginning of the semester and once at the end of the semester. Internal consistency reliability for the total scale was high across groups and time points, while subscale reliability ranged from moderate to high.

As outlined in Table 2, both classes were conducted over 15 weeks and followed weekly routines aligned with their respective syllabi. Table 2 provides an overview of the two instructional procedures, after which each condition is described in greater detail.

Instructional procedures were designed to reflect the core characteristics of each task-sequencing model while maintaining broadly comparable amounts of instructional time and oral practice.

Students completed the FLCAS survey in Week 2 (pre-survey) and Week 14 (post-survey) of the semester. Both surveys were conducted in class, using identical procedures and response formats across both groups. Students were informed that the participation was voluntary and the survey answers would not affect their course grades.

In addition to the FLCAS, students provided basic demographic information, including academic major and year of study. Optional written comments regarding their learning experiences were collected for contextual understanding; however, these comments were not subjected to systematic qualitative analysis and were not treated as primary data sources in the present study.

Quantitative analyses were conducted using SPSS (Version 25). Descriptive statistics (means and standard deviations) were calculated for the FLCAS total score and each subscale at both time points. To examine within-group changes over time (RQ1), paired-samples t-tests were conducted within each instructional condition to examine changes in anxiety from pre- to post-survey.

To compare anxiety change patterns between instructional conditions (RQ2), both gain-score analyses and repeated-measures analyses were employed. Gain scores were calculated by subtracting pre-survey scores from post-survey scores, and independent-samples t-tests with Welch’s correction were used to compare changes between groups. In addition, two-way mixed ANOVAs were conducted with time (pre, post) as a within-subjects factor and instructional condition (RA, W+RA) as a between-subjects factor to examine interaction effects.

To address RQ3, parallel analyses were conducted at the subscale level (CA, TA, FNE) to identify dimension-specific responses to task sequencing. Effect sizes were calculated using Cohen’s dz for within-group comparisons and partial eta squared for ANOVA results.

As shown in Table 3, the two groups were broadly comparable at the pre-survey stage in terms of overall FLCA and subscale means. At the pre-survey stage, the two groups showed similar mean levels of overall FLCA (RA: M = 95.71; W+RA: M = 93.07). Mean scores for CA, TA, and FNE were also broadly comparable across instructional conditions. At post-survey, a reduction in overall FLCA was observed in the RA group (M = 89.86), whereas the W+RA group showed a slight increase (M = 95.49). Subscale-level descriptive patterns indicated that CA and TA decreased in the RA group but increased in the W+RA group, while FNE showed modest decreases in both groups.

Reliability estimates for the total scale were high across groups and time points (α = .92-.94), as presented in Table 4. Subscale reliabilities were generally acceptable, with CA and FNE above .80 and TA showing moderate reliability (α = .65-.77), indicating that the FLCAS functioned reliably in the present instructional context.

To address RQ1, paired-samples t-tests were conducted to examine whether participation in RA-based instruction was associated with changes in FLCA over time. As summarized in Table 5, the RA group demonstrated a significant reduction in total FLCAS scores, t(58) = 2.10, p = .040, dz = -.274. A significant decrease was also observed for CA, t(58) = 2.35, p = .022, dz = -.306. FNE decreased numerically, although the change did not reach statistical significance (p = .065), whereas TA did not significantly change within the RA group.

In contrast, the W+RA group did not show significant reductions in total FLCA. A marginal increase in TA was observed, t(54) = -1.98, p = .053, while other subscales remained statistically stable.

These within-group findings suggest that the RA condition was associated with more favorable anxiety change over time than the W+RA condition, particularly for overall FLCA and CA. Taken together, these results address RQ1 by indicating that participation in the RA-based instructional condition was associated with reductions in overall FLCA and communication apprehension over the semester.

To address RQ2, mixed-design ANOVAs were conducted to determine whether the two instructional conditions showed different anxiety change patterns over time. As shown in Table 6, the clearest between-group difference emerged for TA. A significant Time × Group interaction was observed for TA, F(1, 112) = 6.55, p = .012, partial η² = .055, indicating differential change across instructional conditions. A significant main effect of Group was also found for TA, F(1, 112) = 4.81, p = .030, partial η² = .041.

Interaction effects for CA, F(1, 112) = 3.71, p = .056, partial η² = .032, and total FLCA, F(1, 112) = 3.28, p = .073, partial η² = .028, were suggestive but did not reach conventional levels of statistical significance. No significant interaction was found for FNE. These findings suggest that the two instructional conditions were associated most clearly with different trajectories in test-related anxiety.

These findings indicate that the two instructional conditions were not associated with uniform differences across all dimensions of FLCA. Rather, the most clearly differentiated pattern emerged in the evaluative dimension of anxiety, especially TA. In this respect, the results address RQ2 by showing that the two instructional configurations were associated with different trajectories of anxiety change, most clearly in TA.

The clearest comparative evidence in the present study therefore concerned TA rather than overall FLCA, suggesting that the affective consequences of instructional design were most visible in the evaluative dimension of anxiety.

As a supplementary analysis for RQ2, gain-score comparisons were conducted to clarify the magnitude and direction of between-group change. As presented in Table 7, gain-score comparisons supported the ANOVA findings. A significant between-group difference was observed for TA, t = -2.54, p = .012, d = -.480, indicating greater reduction in TA in the RA group relative to the W+RA group. Gain differences for CA and total FLCA approached significance, whereas FNE showed no differential change.

These results reinforce the RQ2 findings by showing that the clearest between-group difference emerged for TA. In other words, the gain-score analysis confirms that the RA and W+RA conditions were associated with meaningfully different patterns of change, even though those differences were not equally strong across all subscales.

To address RQ3, subscale-level analyses were conducted to identify which dimensions of FLCA were most sensitive to differences in instructional sequencing. As shown in Figure 1, the most pronounced divergence between the two instructional conditions was observed in TA. The RA group showed a slight decline in TA over time, whereas the W+RA group showed an increase, a pattern that is consistent with the significant Time × Group interaction reported in Table 6 and the gain-score difference shown in Table 7.

Figure 2 provides a broader visual comparison of change patterns across the FLCA subscales. While CA showed a reduction tendency in the RA group, FNE remained relatively stable across the two instructional conditions. These visual patterns support the statistical results reported in Tables 5 through 7.

Overall, the subscale-level results answer RQ3 by showing that TA was the most sensitive dimension of FLCA in relation to instructional sequencing, whereas FNE remained comparatively stable across conditions. This pattern suggests that the affective consequences of instructional design may be more visible in some dimensions of anxiety than in others.

Taken together, the results suggest that the most robust group difference was observed for TA rather than for overall FLCA. While the RA group showed significant within-group reductions in overall FLCA and CA, the clearest comparative finding was that the two instructional conditions were associated with different change patterns in the evaluative dimension of anxiety. This pattern suggests that instructional design may matter less for global anxiety reduction than for how evaluative pressure is experienced within a particular oral task configuration.

The reduction in CA within the RA group is broadly consistent with research on task repetition and oral fluency development (Bygate, 2001; Gorsuch & Taguchi, 2010). In the RA condition, learners repeatedly engaged in oral delivery without having to generate original spoken content in real time. This repeated exposure may have reduced uncertainty associated with classroom speaking and gradually increased comfort with oral delivery. Because the linguistic content was externally provided, learners may also have been able to allocate attention more directly to articulation and delivery, thereby reducing some aspects of communication-related burden. At the same time, the absence of a statistically significant Time × Group interaction for CA suggests that this pattern should be interpreted more cautiously than the finding for TA.

The most differentiated pattern emerged for the TA subscale. One possible interpretation is that the two instructional conditions framed oral performance differently in terms of preparation and evaluation. In the RA condition, repeated engagement with a stable oral task format may have supported familiarity and reduced the sense of being tested during classroom performance. In the W+RA condition, the added demands of written preparation, combined with more performance-oriented speaking topics, may have heightened learners’ attention to accuracy, completeness, or self-presentation. In this sense, written preparation may have functioned not only as a planning aid but also as a source of increased evaluative salience within that instructional configuration. However, this interpretation should be made cautiously, because the two conditions differed not only in preparatory sequence but also in task genre and instructional materials. Rather than concluding that writing itself intensifies anxiety, the findings suggest that different instructional configurations may shape the evaluative dimension of classroom anxiety in different ways.

FNE remained relatively stable across both instructional conditions. Unlike CA and TA, FNE did not show significant interaction or within-group changes. This suggests that concerns about peer judgment and classroom evaluation may be less sensitive to task sequencing alone and may depend more on broader classroom climate, interpersonal dynamics, or feedback practices. In other words, some dimensions of anxiety may be more resistant to short-term instructional variation than others.

From a pedagogical perspective, RA-based instruction may provide a relatively low-pressure entry point for oral practice in anxiety-sensitive contexts. However, adding a writing stage does not automatically produce parallel affective benefits. The present findings suggest that the consequences of preparation depend on how preparatory demands, task genre, and evaluative framing are combined in classroom practice. For instructors, this means that the issue is not simply whether learners are given more preparation time. As the results in Tables 5 through 7 indicate, the two instructional conditions showed different change patterns across anxiety dimensions. Although the observed effect sizes were modest, even small shifts in anxiety may be pedagogically meaningful in oral EFL classrooms where affective barriers can influence participation and task engagement over time.

From a general education perspective, these findings are meaningful because university general English courses typically serve students from diverse majors and proficiency backgrounds. In such settings, effective course design should consider not only linguistic practice but also whether particular task configurations support accessible participation and emotionally sustainable engagement.

Several limitations should also be acknowledged. First, the study employed a quasi-experimental design using intact classes, which limits causal interpretation. Because students were not randomly assigned to conditions, pre-existing differences between classes cannot be fully ruled out. In addition, because students self-selected into elective courses, differences in task preference, perceived oral confidence, or prior attitudes toward speaking-related activities may also have influenced the observed patterns. Second, the two instructional conditions differed not only in the presence or absence of a writing stage but also in task materials, topic types, and discourse orientation. Thus, the findings should be interpreted as differences between two naturally occurring instructional configurations rather than as evidence of the isolated effect of writing alone. Third, the study relied on self-report survey data, which capture learners’ perceived anxiety but not necessarily moment-to-moment fluctuations during task performance. Finally, although the total FLCAS demonstrated high reliability, the reliability of some subscales, particularly TA, was more modest; subscale interpretations should therefore be made with appropriate caution.