Review Reports

Comments 1: The introduction is conceptually rich but overly general. It does not sufficiently establish the specific research gap in STEM language learning nor clarify how this study advances. The authors should explicitly anchor the problem within recent empirical findings in AI-enhanced language learning and STEM literacy research. İt is not enough argument for problem statement.

Response 1: Thank you for your detailed feedback and for highlighting the importance of introduction. We carefully reviewed and explicitly grounded the problem statement in recent empirical studies on AI-enhanced language learning and STEM literacy, and more clearly demonstrated how the present study advanced existing research by addressing limitations in current STEM-oriented language learning models. made the necessary adjustments to ensure uniformity in terms of formatting, layout, and style. These changes have been incorporated into the revised version of the manuscript. We appreciate your suggestion and are confident that explicitly linking the identified gap to the proposed framework and empirical investigation, thereby strengthening the justification and originality of the research (pages 1-3).

Comments 2: The stated aim is broad and descriptive. It should be reformulated into clear research questions or testable hypotheses, aligning the framework, digital platform, and

pedagogical experiment within a coherent theoretical chain.

Response 2: We thank the reviewer for this comment. We acknowledge that the stated aim was initially broad and descriptive. In the revised manuscript, we have reformulated the aim into clearly defined research questions and testable hypotheses, establishing a coherent theoretical link between the proposed framework, the digital platform, and the pedagogical experiment (page 2).

Comments 3: “Dialectical logic” is not a recognised methodological framework in contemporary STEM or language-education research. The authors should adopt and report transparent, replicable research design terminology, e.g., quasi-experimental design, design-based research, or mixed-methods.

Response 3: We thank the reviewer for this comment. In the revised manuscript, we have removed the term “dialectical logic” and replaced it with transparent and widely accepted research design terminology, explicitly defining the study as a mixed-methods, quasi-experimental pedagogical investigation (page 9).

Comments 4: ADDIE can guide instructional design but does not constitute a full research methodology. The authors should specify, participants, procedures, data sources, analytic techniques.

Response 4: We thank the reviewer for this comment. In the revised manuscript, we clarify that ADDIE is used solely as an instructional design model, not as a research methodology. We have explicitly specified the research design, participant characteristics, experimental procedures, data sources, and analytic techniques, including quantitative analysis conducted using SPSS, to ensure methodological transparency and replicability (page 9).

Comments 5: Snowball sampling is problematic in an institutional context where the entire student population is accessible. It weakens internal validity. The authors should justify this choice or replace it with cluster sampling or intact group assignment (typical in educational experiments).

Response 5: Thank you for this valuable comment. We agree that snowball sampling is not optimal in an institutional educational context where the student population is accessible, and we acknowledge that its use may raise concerns regarding internal validity.

To address this issue, the description of the sampling procedure has been revised. The study employed an intact group (cluster) assignment design, which is standard practice in educational and classroom-based research. Thirteen naturally existing first-year student groups from the Mining Faculty were included in the study and assigned to experimental and control conditions based on their curricular schedules rather than individual randomisation (page 23).

Comments 6: The manuscript presents descriptive improvements but no inferential statistics, making it impossible to evaluate significance or effect magnitude. The authors must add statistical tests and effect sizes.

Response 6: We acknowledge that the absence of inferential statistics limits the evaluation of significance and effect magnitude. To address this, we included appropriate statistical analyses along with effect sizes to provide a more rigorous and interpretable assessment of the results (page 26).

Comments 7: The discussion reads as confirmatory and promotional rather than analytical. It should: 

• Compare findings with recent AI-in-STEM-language-learning research, 

• Address limitations (sample bias, measurement validity, lack of randomisation), 

• Engage with alternative explanations for learning gains.

Response 7: We acknowledge that the discussion may appear confirmatory. We revised it to include a comparison of our findings with recent research on AI in STEM language learning, explicitly address study limitations (such as sample bias, measurement validity, and lack of randomization), and consider alternative explanations for the observed learning gains (page 26).

Comments 8: Conclusions are overly general and do not align with specific research questions or quantifiable outcomes. Recommendations for educators and implications for STEM curriculum design should be added.

Response 8: We acknowledge that the conclusions are currently presented in a general manner. We revised them to explicitly align with the research questions and the quantifiable outcomes of the study, and we had incorporated evidence-based recommendations for educators along with implications for STEM curriculum design (page 38-39).

Comments 9: Major Revision Needed: All APA components (citations + reference list) must be rewritten according to APA 7 guidelines before publication. The manuscript does not comply with APA 7 standards in either in-text citations or the reference list. In-text citations show major inconsistencies, including the use of numeric citation formats, missing publication years, incorrect use of et al., and absence of page numbers for direct quotations.

The reference list also contains numerous formatting errors: missing years, missing publishers, non-italicized titles, inconsistent punctuation, lack of DOIs, and incorrect ordering. Several in-text citations do not correspond to entries in the reference list.

Response 9: We acknowledge that the manuscript did not fully comply with APA 7 guidelines. We thoroughly revised all in-text citations and the reference list to ensure consistency with APA 7 standards, including correct use of author-date formats, proper use of et al., inclusion of page numbers for direct quotations, accurate formatting of titles and publishers, consistent punctuation, inclusion of DOIs where applicable, and alignment between in-text citations and reference list entries.

Comments 1: Explain sampling and allocation: explain the procedure for allocation to experimental/control subjects (randomization? cluster allocation?). If non-random, change the language of the causal claim to quasi-experimental and discuss limitations.

Response 1: Thank you for this valuable comment. The study employed an intact group (cluster) assignment design, which is standard practice in educational and classroom-based research. Thirteen naturally existing first-year student groups from the Mining Faculty were included in the study and assigned to experimental and control conditions based on their curricular schedules rather than individual randomisation (page 23).

Comments 2: Add a between-group comparison analysis: conduct a statistical analysis that tests for differences in change (post-pre) between experimental and control subjects—report the test statistic, p-value, effect size, and CI. If the data are clustered, use an appropriate test or multilevel model. Do not rely on a correlation of r=0.93 for effectiveness claims.

Response 2: Thank you for your constructive feedback regarding the references We revised it to include a comparison of our findings with recent research on AI in STEM language learning, explicitly address study limitations (such as sample bias, measurement validity, and lack of randomization), and consider alternative explanations for the observed learning gains (page 26).

Comments 3: Strengthen the description of the instrument and scoring: include sample items or instrument appendices, assessment rubrics, training raters, and evidence of content/construct validity. Report inter-rater reliability for open-ended assignment assessments.

Response 3: We acknowledge the need to strengthen the description of the instrument and scoring. We strengthened the description of the instrument and scoring by including the passage, providing assessment rubrics, describing the training of raters, and presenting evidence of content and construct validity. We reported inter-rater reliability for open-ended assignment assessments to ensure transparency and rigor in the evaluation process (page 21).

Comments 4: Clarify the duration/intervention and fidelity: how long (hours) each module is, whether there is variation between groups, instructor engagement levels, platform usage outside of class, etc. Include usage log data to demonstrate exposure (time-on-task).

Response 4: Thank you for this valuable comment. We clarified the duration and fidelity of the intervention by specifying the number of hours for each module, detailing any variations between groups, and describing instructor engagement levels. We included platform usage data outside of class, including usage logs, to demonstrate participants’ exposure and time-on-task (page 31).

Comments 1: Section 2. Materials and Methods contains only some very generic terms, which most likely should be just left out or merged into the introduction. The sole thing that is actually would fit the Methods section is the last paragraph about using SPSS software but it is very short. This could actually be just placed right at the beginning of the section presenting the statistical results (7. Discussion). Nonetheless a reference to the software package itself would be welcome, and also a little more detailed description about how exactly was the software used to obtain the statistics.

Response 1: We thank the reviewer for this insightful comment. We have carefully revised Section “Materials and Methods”. To address your concerns:

1) The generic terms that were previously in the Materials and Methods section have been removed, as suggested.

2) The description of SPSS usage has been expanded and moved to the beginning of the statistical results section (Section 6, Discussion), to provide better context for the analyses.

3) We have added a proper reference to the SPSS software package:

4) We also provided more detailed information on how the software was used, including the types of statistical tests performed (e.g., descriptive statistics, Pearson correlation coefficient) and how the data were processed, to ensure reproducibility.

We believe these changes improve the clarity and appropriateness of the Methods description in our manuscript. (pages 9, 33)

Comments 2: Section 3 is named Results, but it does not contain any actual results, but more or less is a Literature Review. This could actually be renamed as such and placed after the Introduction skipping the generalities.

Response 2: We thank the reviewer for this important observation. Regarding Section 3, which was previously titled “Results,” we agree with the reviewer’s observation. This section has been renamed “Literature Review” to accurately reflect its content and has been moved to follow the Introduction. Redundant generalities have been removed to streamline the text and improve clarity.

Comments 3:

After this literature review the actual description of the framework can commence. However the section 4, which is titled Framework of scientific and linguistic training in the field of stem (here please also use uppercase letters for the acronym STEM) still contains only introductory material, it enumerates some generic principles that should be used in such a framework, but the wording is very confusing and fails to connect naturally with section 5, which appears to be about the actual implementation. But section 5 also fails to refer back to the generic terms described in section 4, just as if it would be two different papers with different scopes.

These two sections should be merged together and have a more natural flow of narration, referencing back to the earlier Literature review section it should then present the main principles that should guide such systems and then continue with implementation details that underline all the enumerated approaches (that are currently presented in section 4).

Response 3: We thank the reviewer for this detailed observation. We fully agree with the concern regarding the flow and connection between Sections 4 and 5. In response, we have:

1) Merged Sections 4 and 5 into a single coherent section, now titled Framework of scientific and linguistic training and implementation of AI-enhanced digital platform for STEM language learning, with the acronym in uppercase as suggested.

2) Restructured the narrative to flow naturally from the Literature Review, presenting first the main principles that should guide such training systems.

3) Ensured explicit references back to the Literature Review and earlier principles throughout the section, so that the framework and implementation are fully interconnected. (page 10)

We believe these changes significantly improve the clarity, coherence, and logical progression of the manuscript.

Comments 4:

Section 6 again starts with a remark that a framework was implemented breaking the flow of reading the paper. Implementation details should be in the previous section, and this section should contain the experimental results. Just don't break the natural flow, and start by stating that the framework was used in the experimental groups of 122 students, then a little bit about the study flow (if necessary) and then continue with the results. The results containing the point values should at the beginning also state how those points were obtained, what scales were used, since gradings can differ from country to country, and not every reader can infer from the tables what the results mean. The columns that show percentage values, even if it is stated in text, should also have explicitly the percentage sign (%) next to the numbers, to allow faster grasping the magnitude of results.

Response 4: We thank the reviewer for this detailed and constructive comment. We fully agree with the points raised and have revised Section 6 accordingly:

Implementation details of the framework have been moved to the preceding section, so that Section 6 now focuses solely on the experimental results.

The section begins by stating that the framework was applied to the experimental groups of 122 students, followed by a brief description of the study flow. (page 23)

Comments 5:

Section 7, the Discussions also again breaks the flow the of reading. Since the framework is already presented it should not state again the key aspects of the framework but go directly to discussing the results from the previous section. It can start for example with short introduction about using SPSS (with concrete details about the configuration, the data that was fed into it and the functions that were used) and then the statistical results. The methods and tests that were used (like the Kolmogorov-Smirnov normality, the Wilcoxon signed-rank T test, etc) should all be referenced and maybe also formulas used should be included here. But definitely needs references so readers can look up the background of those tests.

Response 5: We thank the reviewer for this valuable comment and fully agree. Section “Discussion” has been revised to focus directly on discussing the experimental results, without repeating the key aspects of the framework. A brief introduction to the use of SPSS has been added, including details on the configuration, the datasets used, and the specific functions applied. All statistical methods employed, such as the Kolmogorov–Smirnov normality test and the Wilcoxon signed-rank test, are now explicitly described, referenced. These changes improve the clarity, flow, and reproducibility of the Discussion section. (page 33)

Comments 6:

The tables in section 7 should all be reworked, right now they are very badly formatted, there appears to be some footnotes to it, but it's a mixture of using superscript characters or normal characters (especially normal size characters after numbers, like -7.086c is very confusing. I would recommend to use a single table (or two tables) exactly as it is in the Google document linked in the paper (https://docs.google.com/spreadsheets/d/1NxH1pHC2-b3zuTfFoBOHrnVtxZM1MSG0lAm-cDCUzOw/edit?gid=0#gid=0) and not separate tables for every groups. The joint table looks better and is more conclusive on the results. Also, if you put in the table directly, then there is no need to put a link with the google document. For linking datasets some better data storage options (including from MDPI) can be used. Also please make sure that decimal points are used consequently in the document. Right now, especially in these split tables you have a mixture of using either commas or dots for decimal indicator which is very confusing. I think you should adhere to using only dots and not commas.

Response 6: We thank the reviewer for this detailed comment and fully agree. Footnotes and inconsistent characters have been removed, and all decimal points are now presented consistently using dots only. (pages 33-39)

Comments 7:

Some additional comments on some specific items:

line 661: the link http://stem.ai-systems.kz/profiles.html was not working at the time of the review, please make sure that you only include active links in the paper, that have some guarantee to work in the foreseeable future

main text should contain exact references to tables, for example on line 860 state The table below ...  this should be replaced with reference to actual table number (Table 1 in this case) since tables may get moved around in the final edited document

Table 1 should contain a column specifying what each row refers to, I'm guessing there would be 6 rows with the experimental groups and the 7th row the average values, but this should be clearly expressed

Tables should be made more uniform looking (with a more compact arrangement), currently tables have varying widths on the page (wide tables should take up the whole page width for a better fit) and varying cell heights (at least for the cells containing numbers it would be nicer to have a uniform look)

Lines 945-946 seems to be left from the template document

Please give a reference to the Pearson correlation coefficient introduced on line 943

The sentence in line 975 is unfinished. It should be the sentence that connects the results from the previous section to the discussions section, but it just breaks read flow.

On line 303 you are referring to a paper with two different transliteration of the author name (Grinevich and Hrynevych) please use only one form (preferably the one that the author was using in the original paper. Actually, on all text references when you include the author name in the text you should only put the year in parenthesis and not the full parenthetical citation, see the APA style guidelines regarding narrative in text citations. So, in this case the line would read: "A study by Hrynevych et al. (2023) explores how ...".

Response 7: We thank the reviewer for these detailed and constructive comments and fully agree. The inactive link has been removed. All main text references to tables now refer explicitly to the correct table numbers, ensuring clarity if tables are moved during final editing. Table 1 has been revised to include a clear column specifying each row, distinguishing experimental groups and average values. Lines 945-946 from the template have been removed. A proper reference to the Pearson correlation coefficient has been added on line 943. The unfinished sentence on line 975 has been removed to ensure smooth transition from results to discussion. Finally, all author names in text citations have been standardized according to APA narrative style. These revisions improve clarity, formatting, and adherence to style guidelines throughout the manuscript.

Comments 1: I would like to thank the authors for their careful and comprehensive revision of the manuscript. The revised version clearly demonstrates that the reviewer comments have been addressed in a systematic and thoughtful manner. The explanations provided in the methodology section regarding the research design, implementation process, and analytical procedures have reached an adequate level, thereby enhancing the reliability and replicability of the study. The presentation of the digital platform, its modules, and the AI-supported features is clear and well-structured, enabling readers to clearly understand both the pedagogical approach and the practical implementation. In addition, the discussion and limitations sections have been appropriately revised, indicating that the authors have adopted a critical perspective on their findings and are aware of the methodological boundaries of the study. The quality of the English language is clear, fluent, and appropriate for academic publication.

Response 1: We sincerely thank the reviewer for the positive and encouraging evaluation of the revised manuscript. We highly appreciate your recognition of the systematic manner in which the reviewer comments were addressed and your acknowledgement of the improvements made to the methodology section, research design description, and analytical procedures.
We are particularly grateful for your comments on the clarity and structure of the presentation of the digital platform, its modules, and AI-supported features, as well as for noting the strengthened discussion and limitations sections. Your feedback confirms that our efforts to enhance the transparency, reliability, and replicability of the study have been effective.
We also appreciate your remark regarding the quality of the English language, as careful academic language editing was an important part of the revision process. Thank you once again for your constructive assessment and valuable support, which have significantly contributed to improving the quality of the manuscript.

Comments 1: It would be nice to see a paragraph at the end of the Introduction pointing out what to expect in the following sections.

Response 1: Thank you for this valuable suggestion. A concluding paragraph has been added to the end of the Introduction outlining the structure of the paper and briefly indicating what readers can expect in the following sections. (p.3)

Comments 2: Also, make sure that the tables in the Discussion section are not split across pages.

Response 2: Thank you for the comment. We have ensured that all tables in the Discussion section are now formatted so that they are not split across pages. (pp.33-39)