Augmented Reality’s Impact on English Vocabulary and Content Acquisition in the CLIL Classroom

Featured Application

An AR-enhanced CLIL (Content and Language Integrated Learning) unit for secondary EFL (English as a Foreign language) Science uses markerless, mobile AR to overlay 3D models and just-in-time English vocabulary on standard classroom tasks. The unit is preconfigured and runs on students’ phones, supporting vocabulary and content acquisition in ordinary classrooms. The approach was evaluated in a classroom study where one section used AR and a parallel section used print materials.

Abstract

This study interrogates whether Augmented Reality (AR) enhances vocabulary and content acquisition within Content and Language Integrated Learning (CLIL), situating the question in the broader debate on how immersive, multimodal technologies shape achievement and engagement. This study’s novelty lies in its direct AR-versus-print comparison in a real CLIL classroom using markerless, smartphone-based technology. Using a mixed-methods, classroom-based experiment, we drew on a convenience sample of 129 secondary students (ages 16–18), assigning them to an AR intervention (n = 64) or a print-based control (n = 65). Both cohorts received parallel instruction covering identical objectives and materials; vocabulary attainment was gauged using matched pretest and post-test measures, while engagement, attitudes, and perceived usefulness were captured through paired pre- and post-surveys and open-ended prompts. Quantitative analyses compared change scores across conditions and were complemented by qualitative summaries of learner comments. Results indicate that exposure to AR exerted a positive influence on learners’ engagement and supported learning processes, with perceptible shifts in students’ views of AR between baseline and post-intervention; nevertheless, effects were heterogeneous across instruments, items, and subgroups, suggesting that benefits accrued in a targeted rather than uniform fashion. Compared to the print-based group, students using AR demonstrated greater gains on visually supported vocabulary and content items, while other items showed no significant differences between groups. We conclude that AR constitutes a promising pedagogical resource for CLIL, capable of scaffolding vocabulary/content development and motivating participation, while the observed variability underscores the need for principled, context-sensitive integration. Future work should specify boundary conditions—such as task type, prior proficiency, cognitive load, and technology familiarity—and employ robust mixed-methods designs to determine for whom, and under which instructional circumstances, AR yields the greatest and most sustainable gains.

1. Introduction

Educators and policymakers increasingly ask whether emerging classroom technologies demonstrably improve students’ learning processes, not merely their curiosity [1]. In Content and Language Integrated Learning (CLIL) settings, where subject knowledge and language are learned together, Augmented Reality (AR) is often proposed as a means to visualize abstract content and provide immediate lexical support [2,3,4]. Yet classroom-based evidence on AR’s actual impact on learning—especially in ordinary secondary CLIL contexts—remains comparatively scarce, which motivates the present study and underscores the significance of our positive findings. Converging reviews further indicate that AR/VR commonly leverage immersive, interaction-rich—often game-based—experiences that boost motivation and learning outcomes in language education [2,5,6,7], with games emerging as a particularly popular and effective format due to their emphasis on interaction [2].

The purpose of this research is to gather new, classroom data on a technology that may change how English content and vocabulary are taught and learned in CLIL. We analyze students’ vocabulary and content learning and their motivation/perceptions during a short, markerless, mobile AR intervention delivered with a standardized workflow (Zapworks Designer SDK) in a regular secondary classroom—conditions chosen to maximize ecological validity and teacher uptake. The present study was designed to examine whether AR can support vocabulary and content learning under realistic classroom conditions while remaining feasible for teachers to implement. Our specific aim was to compare AR and print-based delivery of identical content and tasks to determine their effects on student learning and perceptions. While our results confirm some well-established benefits of AR, the study adds value by demonstrating how AR can be integrated into regular secondary CLIL lessons using common smartphones, providing practical evidence for teachers and schools seeking scalable and realistic implementations.

In line with recent syntheses, we frame the design to support self-directed pacing and mobile use—features frequently reported in current studies—and to facilitate potential outdoor/experiential activities [2]. We also acknowledge the need for teachers to design aligned materials that capitalize on AR affordances rather than entertainment alone, so that learning objectives remain central [2]. References were identified through a targeted search of peer-reviewed articles published in the past ten years, focusing on augmented reality, CLIL, and vocabulary acquisition. Selection criteria included relevance to language education, methodological rigor, and empirical evidence of AR effectiveness in similar educational contexts.

The state of the field shows growing interest and encouraging results, with several recent studies by Belda-Medina and colleagues reporting gains in vocabulary and learner motivation, as well as practical guidance for teacher preparation and authoring in EFL/CLIL contexts [8,9]. These works—together with broader syntheses—suggest AR’s potential to foster interactive, engaging environments that benefit language and content learning. At the same time, implementation quality and teacher readiness remain pivotal for realizing these benefits in everyday schools. Consistently, prior reviews highlight that the most recurrent effects of AR/VR are improvements in learning achievement and motivation [5,6,10], and that AR supports immersive learning and active interaction [7].

There are also diverging views that matter for adoption. Huertas-Abril et al. [11] stress the necessity of explicit teacher training for effective AR integration, echoing reports of limited pedagogical/technological readiness; our context review also finds limited evidence linking AR to teacher development. Although most students find AR easy and intuitive, a noticeable minority still prefer traditional materials, and many are lukewarm about using AR beyond the study. These patterns argue for blended designs and strategies that sustain engagement over time. At the same time, documented constraints—such as potential distraction by virtual content, added classroom-management demands, and time-consuming implementation—must be managed through careful task design and guidance [2,12,13,14,15]. Ensuring that AR activities are tightly aligned with curricular goals helps prevent gameplay from overshadowing learning targets [2].

This article examines AR’s effects on English vocabulary and content acquisition and on students’ motivation and perceptions in a secondary CLIL unit. Anticipating the results, AR proved feasible in ordinary classrooms, enhanced students’ motivation, and contributed to learning—students using AR showed better retention and understanding of vocabulary and content than those taught with traditional materials—while some preferences for conventional resources persisted, reinforcing the value of blended approaches and teacher preparation [1,2,3,4,8,9]. This expectation accords with recent evidence that AR can increase engagement and motivation, reduce anxiety, and bolster satisfaction and achievement—particularly in vocabulary and reading-related outcomes—while also supporting knowledge retention [16,17,18,19,20,21]. This study adds to AR–CLIL research by testing a markerless, smartphone-based design under real classroom conditions and directly comparing AR and print-based delivery using identical content and tasks.

2. Materials and Methods

2.1. Study Design and Setting

We conducted a classroom-embedded, sequential mixed-methods study with two parallel groups. Students completed the same CLIL content and vocabulary targets delivered in two modalities: an Augmented Reality (AR) lesson (Experimental Group, EG) or an equivalent print handbook (Control Group, CG). The design isolates delivery mode as the only manipulated variable while holding content, timing, and tasks constant.

2.2. Participants and Context

The sample comprised 129 students from a Spanish high school who study English as a foreign language (EFL). Average proficiency was CEFR B1. Students were assigned to two groups—CG and EG—via cluster randomization assignment (simple allocation without prior stratification). Entire intact classes were treated as clusters and randomly assigned to either condition, with no reassignment of individual students. The study analyses AR’s impact on content and vocabulary learning and documents positive and negative aspects of AR use observed in this setting.

2.3. Materials

AR intervention (EG). The EG completed a 30 min session using 3D SDK-based AR teaching units. Tasks presented the same target content and vocabulary as the CG, with on-screen prompts guiding learners through activities. Illustrative screenshots are provided in the Appendix B (Figure A1, Figure A2, Figure A3 and Figure A4). Students used their own mobile devices (smartphone) in a standard classroom environment. The AR module was built using Zapworks SDK, which operates on common smartphones with basic internet access. Teachers only needed a brief orientation to run the system. Licenses for the software were covered by our publicly funded research project and will be credited accordingly, ensuring there were no additional costs to the school or participants.

Control Materials (CG). The CG received an activity handbook containing exact replicas of the EG activities adapted to a traditional print format. Session duration was also 30 min (Appendix B Figure A5, Figure A6, Figure A7, Figure A8 and Figure A9). In future iterations, assessment instruments will be refined to group items by class and complexity, allowing for a clearer understanding of which types of vocabulary and content benefit most from AR.

2.4. Instruments

Baseline survey (demographics and tech attitudes). A pre-survey collected gender, age, and other descriptors to characterize the participant corpus. Technology-related attitudes were analyzed with the ARAAS scale (as used in prior exploratory work), to contextualize students’ baseline familiarity and dispositions. In addition, a separate linguistic pre-test was administered to assess students’ initial vocabulary and content knowledge for the CLIL unit. This allowed us to compare groups before the intervention and confirm that no significant baseline differences existed.

Pre-test (baseline). A pre-test captured students’ familiarity with AR and classroom technology to ensure participants approached the experiment without prior bias toward the treatment modality.

Immediate post-tests. Right after the session, the EG completed a post-test covering difficulty and accessibility of AR and vocabulary/content learned. The CG completed a parallel post-test aligned to the print workflow to enable comparative analysis across formats.

Post-questionnaire (immediate). In addition to the performance post-test, both groups completed a post-questionnaire measuring ease of use, engagement, device convenience, and willingness to continue using the instructional format. The questionnaire included 18 items: Items 1–14 referred to the experimental group’s experience with AR, while Items 15–18 were parallel questions for the control group, which completed the same tasks using printed materials. Duplicate wording was intentional to allow direct comparison between groups.

Delayed follow-up (≈30 days). A post-survey 30 days after the experiment assessed students’ perception and willingness to continue using AR, capturing attitudes/usability at ≈30 days; no robust delayed-learning signal was detected.

Operationalization of vocabulary/content learning. Learning was assessed with both productive and receptive items, which were identical across modalities; only the delivery format differed.

• Productive vocabulary: fill-in-the-gap exercises plus picture–word matching.
• Receptive vocabulary: multiple-choice questions.

2.5. Procedure

Figure 1 illustrates the sequential procedure followed during the experiment. The process was divided into five stages, starting with the collection of baseline data and progressing through the intervention, post-tests, and follow-up assessments. Students in the two groups completed the intervention separately, with different classrooms and time slots used to prevent cross-group contamination or sharing of materials between experimental and control clusters. All sessions were teacher-supervised and followed a standard protocol: device readiness was checked at the start, clear rules were given to prevent unrelated use, and tasks proceeded in a fixed 30 min sequence.

2.6. Data Analysis

Quantitative data were summarized with descriptive statistics (means, SDs, proportions). To test learning effects, we compared post-test outcomes between EG and CG and examined within-group changes (pre–post for attitudes, where applicable). Given the item-level and ordinal nature of several measures, nonparametric tests were planned (e.g., Wilcoxon signed-rank for paired changes; Mann–Whitney U for between-group comparisons), reporting Z statistics, exact/asymptotic p-values (α = 0.05), and effect sizes (r) calculated using the formula r = Z/√N to provide a better understanding of the magnitude of the observed effects. In this formula, N refers to the total number of observations included in the test—either the number of paired cases for within-group comparisons or the combined sample size of both groups for between-group comparisons. Missing data were handled via pairwise deletion for item-level analyses; sensitivity checks were conducted where appropriate. Descriptive statistics are reported to one decimal place for consistency across tables. Effect sizes (r) were calculated for all non-parametric tests using the formula r = Z/√N, and the direction of differences was noted to clarify which group showed higher performance. A Holm correction was applied to control for the risk of Type I error due to multiple comparisons across the nine test items. Qualitative comments (when present) were used to contextualize quantitative trends.

2.7. Ethics and Transparency

The study was conducted in accordance with the Declaration of Helsinki and followed the institutional ethical standards of the University of Alicante (https://web.ua.es/es/vr-investigacio/comite-etica/, accessed on 22 September 2025). Participation was voluntary, with informed consent obtained from students and/or guardians in line with local policy. No identifiable data were collected or reported, and all responses were anonymized. According to local institutional policy, minimal-risk classroom studies using only anonymized educational data do not require a formal approval code, which is why none was issued in this case. during the AR-mediated tasks. As the research involved only anonymized educational data and minimal risk classroom activities, a specific approval code was not required by the institutional ethics committee. Thus, the research complied fully with institutional and international ethical standards.

2.8. Availability of Materials

To support replication, the following materials will be made available: data tables (instrument specifications, descriptive statistics, and item-level outcomes; packaged as

Table A1. Participant characteristics.

Descriptive Statistics	M	SD
1. Gender	1.5	0.5
2. Language Spoken	1.6	1.3
3. Mother Tongue	2.0	1.9
4. Group	2.1	1.2
5. Age	1.5	0.5
6. General English Level	3.0	1.1
7. Speaking Level	3.0	1.2
8. Listening Level	3.2	1.2
9. Reading Level	3.2	1.2
10. Writing Level	2.9	1.1
11. Type of English Certificate	2.0	2.5

Values are M (SD). Categorical variables were numerically coded; higher codes indicate higher categories. CEFR-based proficiency levels were coded 1 = A1, 2 = A2, 3 = B1, 4 = B2, 5 = C1/C2. n per item reflects valid responses; missing data handled pairwise.

,

Table A2. Descriptive statistics for pre-test items (N varies by item).

Descriptive Statistics	N	M	SD
1. Continue the sentence. “A peak is…”	115	0.89	0.32
2. Continue the sentence. “to give up is…”	115	0.87	0.34
3. Continue the sentence. “A predator is…”	115	0.90	0.30
4. Continue the sentence. “To bring forward is…”	115	0.60	0.49
5. Continue the sentence. “Wilderness can be found in…”	115	0.89	0.32
6. This is a…	115	0.98	0.13
7. This animal is…	115	0.98	0.13
8. This picture represents…	115	0.29	0.45
9. This picture represents…	115	0.54	0.50

Note. Pre-test items were coded 1 = correct and 0 = incorrect. The Mean therefore represents item difficulty as the proportion of correct responses (higher Mean = easier item). The Standard Deviation reflects variability expected for dichotomous items [√p(1 − p)]. N indicates the number of valid responses per item. Differences between per-item N (115) and the total sample (N = 129) reflect occasional missing responses; analyses used pairwise deletion.

,

Table A3. Participants’ technological affinities and digital use.

Descriptive Statistics	M	SD
1. Ownership of their own devices	2.8	0.4
2. Age 1st smartphone	2.9	0.8
3. TikTok frequency	3.9	1.4
4. Instagram frequency	4.1	1.1
5. Snapchat frequency	1.2	0.6
6. Twitter frequency	2.1	1.3
7. Tumblr frequency	1.0	0.4
8. Facebook frequency	1.1	0.4
9. YouTube frequency	3.6	1.0
10. Pinterest frequency	2.2	1.2
11. Kik frequence	1.0	0.1
12. Discord frequence	2.1	1.2
13. Others frequence	1.7	1.3
14. Videogames	2.9	1.2
15. Smartphone usage	3.3	1.1

Values are M (SD). Frequency items use a 5-point scale (1 = Never, 5 = Always). “Age 1st smartphone” is a 5-point ordinal bin (lower values = younger acquisition). n per item excludes missing responses (pairwise deletion).

,

Table A4. AR attitudes (pre).

Items	M	SD
1. AR improves performance	3.4	0.9
2. AR improves academic productivity	3.4	0.9
3. AR makes learning easier	3.6	0.8
4. AR is easy to use	3.4	1.0
5. Learning how to use AR is easy	3.6	0.9
6. It is easy to become SKILLFUL at AR	3.4	0.9
7. Studying with AR is a good idea	3.5	0.9
8. Studying with AR is a wise idea	3.3	1.0
9. Students are positive towards AR	3.6	0.9
10. Students intend learning more about AR	3.3	1.0
11. Students intend to be frequent users of AR	2.8	0.9
12. Students have confidence to learn via AR	3.3	0.9
13. Students have necessary skills for using AR	3.2	0.1
14. Students enjoy AR based on their own experience	3.4	0.9
15. Students believe AR knowledge is needed for future jobs	3.2	1.1
16. Students have no difficulty accessing and using AR	3.1	1.0

Values are M (SD). Scale anchors: 1 = Strongly disagree, 5 = Strongly agree. n per analysis excludes missing data (pairwise deletion).

,

Table A5. AR attitudes (post).

Descriptive Statistics	M	SD
1. How easy was using AR during the experiment for you? [How was using AR technology?]	3.89	0.8
2. How easy was using AR during the experiment for you? [How engaging was AR applied to the experiment?]	3.38	0.7
3. How easy was using AR during the experiment for you? [How interesting was the use of AR applied to experiment?]	3.3	0.9
4. How easy was using AR during the experiment for you? [How intuitive was using AR applied to the experiment?]	3.8	0.8
5. How easy was using AR during the experiment for you? [How problematic was following the steps during the experiment?]	3.4	0.9
6. Is it likely you will try AR again after the experiment for your own interest?	2.9	1.3
7. How convenient is using each of these devices for learning with AR? [smartphones]	3.5	1.1
8. How convenient is using each of these devices for learning with AR? [tablets]	3.7	1.1
9. How convenient is using each of these devices for learning with AR? [laptops]	3.4	1.1
10. How do you feel about the use of the following devices in your classes for EDUCATIONAL purposes after having had the experience during the experiment? [smartphone]	3.5	1.0
11. How do you feel about the use of the following devices in your classes for EDUCATIONAL purposes after having had the experience during the experiment? [tablet]	3.7	1.1
12. How do you feel about the use of the following devices in your classes for EDUCATIONAL purposes after having had the experience during the experiment? [laptop]	3.8	1.3
13. Have you used AR after the experiment?	0.5	0.5
14. How interested are you in using AR technology in your classes after having used it during the experiment?	3.2	1.1
15. How easy was using AR during the experiment for you? [How was using AR technology?]	3.89	0.8
16. How easy was using AR during the experiment for you? [How engaging was AR applied to the experiment?]	3.38	0.7
17. How did you feel after using AR during the experiment?	3.1	1.1
18. How did you feel after using THEORETICAL MATERIALS during the experiment?	3.1	0.7

Values are M (SD). All items use 5-point scales; ease/intuition items are anchored at 1 = Very difficult/not intuitive, 5 = Very easy/very intuitive; engagement/interest items at 1 = Not at all, 5 = Very much; device convenience at 1 = Not convenient, 5 = Very convenient. Higher scores reflect more positive evaluations. n per item excludes missing responses (pairwise deletion).

and

Table A6. Vocabulary and content post-test results for EG and CG.

Language Test Results	Z	p	r (Effect Size)
1. Continue the sentence. “A peak is…”	−3.21	0.0	0.28
2. Continue the sentence. “to give up is…”	0.0	1.0	0.00
3. Continue the sentence. “A predator is…”	−1.5	0.1	0.13
4. Continue the sentence. “To bring forward is	−1.2	0.2	0.10
5. Continue the sentence. “Wilderness can be found in…”	−2.5	0.0	0.22
6. This is a…	−1.4	0.1	0.12
7. This animal is…	0.0	1.0	0.00
8. This picture represents…	−2.6	0.0	0.23
9. This picture represents…	−3.1	0.0	0.27

Note. Z = standardized Mann–Whitney statistic; p = two-tailed p-value; r = effect size calculated as Z/√N. Significant differences are in bold. Negative Z values indicate higher performance by the Experimental Group (AR) compared to the Control Group (Print). For each item, we report Z, p, and effect size r (Z/√N); negative Z indicates higher performance by the AR group and a Holm correction was applied to control for multiple comparisons across the nine test items. Items 1, 5, 8, and 9 showed significant advantages for AR (p < 0.001). Items 2, 3, 4, 6, and 7 were non-significant (ps ≥ 0.10).

in Appendix A), (ii) AR unit screenshots (representative images in Appendix B, Figure A1, Figure A2, Figure A3, Figure A4 and Figure A5), and (iii) the printable handbook in PDF (page captures in Appendix B, Figure A6, Figure A7, Figure A8 and Figure A9).

3. Results

Results are organized as: (3.1) socio-demographic characteristics; (3.2) technological affinity and digital use; (3.3) attitudes toward AR (pre/post); and (3.4) vocabulary and content learning outcomes, (3.5.) Summary of Key Findings. Descriptive statistics are presented as M (SD). Where applicable, non-parametric tests (Wilcoxon signed-rank) are reported with exact Z and p values from the thesis tables.

3.1. Socio-Demographic Characteristics of Participants

A total of 129 secondary students (16–18 years) participated (Experimental Group, EG: n = 64; Control Group, CG: n = 65). The cohort showed a small male representation, and Spanish was the predominant language across the sample. The self-reported English proficiency clustered around the mid-B1 range, with writing rated comparatively lower than other skills. Rates of formal English certificate acquisition were also modest. These descriptive characteristics contextualize the subsequent findings. A comparison of pre-test scores showed no significant differences between the experimental and control groups, indicating baseline equivalence prior to the intervention.

3.2. Technological Affinity and Patterns of Digital Use

Overall, students demonstrated high access to and use of personal devices and platforms. Device ownership was frequent and early adoption of smartphones was common. Social media use was dominated by Instagram and TikTok, with YouTube also widely used. In terms of learning tools, students reported frequent use of smartphones for school tasks, moderate use of laptops, and limited use of tablets. The Technological Affinity table further indicates regular videogame play and the presence of other platforms (e.g., Discord) albeit with lower frequency.

3.3. Attitudes Toward AR Technology (Pre/Post)

Pre-survey. Before the intervention, students reported a moderate level of interest in AR for learning and a moderate intention to learn more about it. The intention to use AR frequently was lower, indicating heterogeneous initial enthusiasm across the cohort.

Post-survey (≈30 days). Post-intervention, perceived ease and intuitiveness of AR remained high. Students rated the convenience of smartphones, tablets, and laptops for AR-supported learning positively. However, willingness to try AR again for personal interest showed a slight decrease relative to pre-intervention levels, and small declines were observed, in general, engagement/interest indicators. Overall, device-related attitudes remained stable, while enthusiasm to re-engage with AR outside the classroom dipped marginally.

Qualitative feedback indicated that while most students found the AR activities intuitive, some reported challenges related to usability, such as difficulties aligning the camera with markers and occasional lag. Others mentioned confusion in more complex tasks, particularly those requiring precise manipulation or spatial interpretation.

3.4. Vocabulary and Content Learning Outcomes

Learning outcomes were assessed through receptive and productive vocabulary tasks aligned with the CLIL unit. A Wilcoxon signed-rank analysis of item-level prompts showed significant pre/post shifts for a subset of vocabulary/content items. Specifically, significant effects were observed for several prompts, including ‘A peak is…’, ‘to give up…’, ‘A predator is…’, ‘Wilderness can be found in…’, and two image-based prompts (‘This picture represents…’), with Z values in the −2.5 to −3.2 range and p < 0.001. Items such as ‘To bring forward…’ and ‘This animal is…’ did not reach statistical significance (p ≥ 0.10). Taken together, these results indicate selective learning gains tied to the target items that were most saliently scaffolded during the AR-mediated tasks. Exact Z, p, and effect size values for each item are reported in Table A6 to provide transparency and to show where the experimental group (AR) significantly outperformed the control group (print). These results show that AR supported learning for certain items, with significant gains in tasks directly scaffolded by the digital overlay, while other tasks remained unchanged or showed no significant differences between groups. Specifically, items such as ‘To bring forward…,’ ‘This animal is…,’ and other low-visual-support prompts did not show significant pre/post improvement, indicating that AR’s benefits were limited to tasks where the visual component directly enhanced understanding. These findings reflect immediate post-test performance only and should not be interpreted as evidence of long-term retention, which was not directly measured.

Retention comparison (EG vs. CG). The experiment within the CLIL classroom demonstrated that students using AR showed higher immediate performance on selected vocabulary and content tasks compared to those who used traditional methods. However, because only immediate post-tests were conducted, these results should not be interpreted as evidence of long-term retention. This improvement highlights AR’s potential to facilitate more effective learning outcomes, a finding supported by a systematic review which noted AR’s benefits in fostering interactive and engaging learning environments [22].

3.5. Summary of Key Findings

(1)

Sample profile. The cohort was balanced across conditions (EG = 64; CG = 65) and linguistically diverse, with Spanish predominant. Overall English proficiency clustered around B1, with writing comparatively lower (≈high A2). These patterns align with links between engagement/knowledge and mobile-AR performance [23] and with diverse certification pathways [24]; gender representation effects are consistent with Gómez-Trigueros and Yáñez de Aldecoa [24].

(2)

Technology access and platforms. Students reported high access and use of smartphones and strong familiarity with mainstream platforms (e.g., Instagram, TikTok, YouTube), while tablets were least used (M = 1.2) and laptops moderate (M = 2.5). Device-level attitudes mirrored prior work showing positive sentiments toward certain tools and in our data smartphones/laptops received the most favorable feedback.

(3)

Perceptions of AR vs. theoretical materials. Post-survey means were similar for AR and theoretical materials (both M = 3.1), but variability was greater for AR (SD = 1.1) than for theoretical materials (SD = 0.7), indicating more divergent experiences with AR. Thus, AR was generally well-received yet more heterogeneous, whereas theoretical materials produced steadier satisfaction.

(4)

Vocabulary/content learning at item level. Item-level tests identified significant effects for several prompts (e.g., ‘A peak is…,’ ‘to give up…,’ ‘A predator is…,’ ‘Wilderness can be found in…,’ ‘This picture represents…’) with associated Z and Sig. values reported. Where prompts did not reach significance (e.g., ‘To bring forward…,’ ‘This animal is…’), the pattern suggests the need for additional scaffolding or extended practice windows in CLIL-AR tasks.

4. Discussion

This study investigated whether integrating augmented reality (AR) into a CLIL unit improves English vocabulary and content acquisition, and it examined learners’ perceptions of AR under regular classroom conditions. The working hypotheses were that (i) AR-enhanced tasks would yield measurable gains on the targeted items and (ii) students would report positive attitudes toward AR’s usability and pedagogical value [1,2,3,9,10].

4.1. Principal Findings and Interpretation

Students improved from pre- to post-test on a small set of questions—especially two that used pictures and one about “wilderness.” Other questions showed little or no change. Thus, AR helped most when the digital overlay actually supported understanding; it added little when the question was already easy or not closely tied to the visuals [4,7].

Students rated AR to be easy/intuitive and judged smartphones/tablets/laptops as convenient for AR-supported tasks in class [2,12,19]. At the same time, the intention to re-use AR for personal interest decreased slightly at the 30-day check, suggesting that one-off exposure—despite being usable—does not automatically translate into sustained self-directed adoption [6,10,21]. In practical terms, AR appears classroom-feasible and pedagogically promising, yet its motivational pull may attenuate unless it is embedded as part of a recurring, purposefully sequenced routine [5,6].

4.2. Relation to Previous Studies and Theoretical Framing

Qualitative feedback further highlighted that usability issues, such as marker alignment difficulties and task-related confusion, influenced how students experienced AR activities. These insights help explain the variability observed across tasks and outcomes. Placed against prior research in AR-supported language learning and CLIL, the present findings converge on two recurrent themes:

(a)

AR can heighten and facilitate learning when its affordances align with the target knowledge [1,7,21];

(b)

Effects are task-dependent rather than uniform [4].

The current item-level pattern strengthens that view by engagement showing, in a secondary CLIL setting, that near-identical vocabulary prompts can diverge depending on how directly AR scaffolds the intended meaning [17,18,20]. From a CLIL perspective, this is consistent with the 4Cs framework: AR best supports the content–communication–cognition nexus when it renders disciplinary meanings perceptible and discussable in English. This also aligns with Cognitive Load Theory, as items directly scaffolded by AR likely reduced extraneous cognitive load, making complex concepts easier to process, while items without clear visual or spatial support may not have benefited to the same degree.

4.3. Pedagogical Implications for CLIL

Three design implications follow:

• Target high-leverage content. Prioritize concepts that profit from spatial/visual cues (e.g., landforms, ecosystems, processes) and make the overlay indispensable (label, compare, or manipulate the object) [4,5].
• Blend and sequence. Interleave brief AR tasks with non-AR consolidation (pre-teach → AR exploration → debrief/practice) to stabilize gains and offset novelty fade [2,6,10].
• Support teachers. Although ease-of-use ratings were high, lightweight training (marker handling, pacing, troubleshooting) is advisable, so classroom time remains cognitively productive [13,14,15,17]. These implications reflect our study’s findings that AR was most effective for visually scaffolded vocabulary and content, aligning with the CLIL 4Cs framework and Cognitive Load Theory by emphasizing meaningful, manageable, and integrated learning experiences.

4.4. Methodological Considerations

The study used intact classes (EG = 64; CG = 65; total N = 129), which supports ecological validity but limits causal claims if baseline equivalence on pre-measures cannot be verified for all variables [4,8]. The intervention window was brief, and analyses centered primarily on item-level outcomes. While this granularity elucidates where AR yields clear benefits, it constrains inferences about broader proficiency gains. An approximately 30-day follow-up survey was administered to examine participants’ perceptions of using AR as a learning tool (attitudes/usability); however, delayed learning effects could not be robustly established (cf. [1]). These factors recommend caution in generalization and emphasize within-study contrasts (see also [17,18,21], on retention and performance). It should be noted that the study involved a single 30 min session and that the 30-day follow-up measured only attitudes and usability rather than knowledge retention; therefore, results should be interpreted with caution, and future research should employ longer interventions and include delayed post-tests to better evaluate sustained learning outcomes.

4.5. Future Research Directions

Subsequent work should extend to multi-site cohorts to test robustness across contexts, adopt longer instructional sequences that interleave AR and non-AR practice to examine retention and transfer, compare AR affordances (marker-based vs. markerless; static labels vs. manipulable 3D; teacher-led vs. student-led exploration) to isolate which features most reliably drive learning and sustained engagement, and incorporate process data to explain why specific items benefit [2,5,6,7,10] (and see [18,20] for cognitive load and task design considerations). Future research should also include teachers’ perspectives on classroom management, preparation demands, and perceived pedagogical value to better inform AR implementation in CLIL settings. Among these directions, developing multi-session interventions and integrating delayed post-tests to assess long-term retention should be prioritized as the next key steps for advancing AR–CLIL research.

5. Conclusions

The experiment within the CLIL classroom showed higher immediate performance on selected vocabulary and content tasks compared to those who used traditional methods. This improvement highlights AR’s potential to facilitate more effective learning outcomes, in line with meta-analytic and review evidence of AR-related learning gains and interaction-driven benefits [1,2,7], and with domain-specific reports of gains in vocabulary and reading-related outcomes [3,21]. Findings on retention and task performance are likewise echoed in controlled studies using classroom-embedded AR [17,18,20].

Technological affinity and user experience with AR were generally positive among students, who found the technology easy and intuitive to use despite some initial challenges. This suggests that with proper introduction and support, students can quickly adapt to AR, enhancing their learning experience [2,12,19]. However, the study also noted that while AR was well-received, some students preferred traditional materials, indicating a need for a blended learning approach that integrates both AR and conventional methods to cater to diverse learning preferences [4,5,6,10]. This aligns with reports that motivation gains can fade without purposeful sequencing and that game-based/interactive designs are most effective when tightly aligned to curricular aims [2,5].

The study’s findings have several implications for future educational practices. Firstly, it is crucial to provide adequate training for educators to successfully integrate AR into the classroom. Teachers need to be proficient in AR technology to effectively facilitate its use and address any issues that might arise [10]. Attention to classroom orchestration and usability—from pacing to troubleshooting—can mitigate distraction and cognitive-load issues reported in device-rich lessons [12,13,14,15]. This integration should be balanced with traditional teaching methods to ensure a comprehensive educational experience [4,5,6]. Secondly, future studies should scale up, lengthen exposure, and compare specific AR affordances to identify robust design levers for durable vocabulary and content learning in CLIL [1,2,7], including teacher-led vs. student-led exploration, marker-based vs. markerless tracking, and levels of interactivity, with attention to cognitive load and retention [17,18,20]. Finally, the growing body of work in EFL/CLIL contexts underscores practical pathways for uptake—authoring workflows, teacher development, and evidence of motivational and lexical gains [8,9].

Future research could explore the use of higher-immersion devices, such as AR glasses or mixed reality headsets, to determine whether increased immersion leads to different or enhanced learning outcomes compared to mobile phones. In addition, further studies could also incorporate a wider range of English tasks, including writing and speaking activities, to assess whether AR can support different language skills beyond vocabulary and content acquisition. The next version of the AR unit will include richer and more contextually meaningful 3D images that are directly linked to abstract vocabulary items, thereby enhancing the connection between visual elements and learning targets.