Mastery of Listening and Reading Vocabulary Levels in Relation to CEFR: Insights into Student Admissions and English as a Medium of Instruction

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This was a a well written and interesting study, and potentially a very useful one. I have e few comments though.

1. The last sentence of the abstract should be better linked to the findings mentioned in the preceding sentences, and perhaps explained slightly better. On my first reading this very important passage was not sufficiently clear/explicit in my opinion.

2. My main problem with this study is the sample. I miss an explanation for why first year and fourth year respondents were selected. Why, and what are possible implications, for instance for the statical analysis as well as the findings? Whether there was any difference between the groups/levels is a question that comes to mind. In my opinion this needs to be better addressed in relation to the validity of the findings.

3. By the way, is it Pearson's or Spearman that is used in the correlation analysis? Would not the latter be most appropriate given the sample?

4. I think that the CEFR construct with its focus on reading, writing, listening and speaking could be explained a bit, it should come as no surprise that it is not always a consistent predictor of academic skills. Indeed, even the TOEFL and IELTS tests are not either. This makes the finding, the utility of supplement testing with appropriate vocabulary tests particularly compelling.

No further

Author Response

Response to Reviewer 1 Comments
1. Summary
Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions in track changes in the re-submitted files.
2. Point-by-point response to Comments and Suggestions for Authors
Comment 1: The last sentence of the abstract should be better linked to the findings mentioned in the preceding sentences, and perhaps explained slightly better. On my first reading this very important passage was not sufficiently clear/explicit in my opinion. Response 1: Thank you for pointing this out. We agree with this comment. Therefore, we have revised the last sentence of the abstract as follows: “Given the positive correlations observed between students’ CEFR levels and their mastery of vocabulary levels, this study’s findings suggest the inclusion of aural/written vocabulary levels as additional indicators for ensuring student academic success in EMI institutions.” And we added two sentences to explain further the findings as follows: “Regression analyses were further conducted to determine the extent to which the combination of receptive aural and written vocabulary knowledge predicts the CEFR levels. The results indicated that the regression model included only UVLT scores better predicted the CEFR levels." This change can be found on page 1, paragraph 1, line 27-34.   Comment 2: My main problem with this study is the sample. I miss an explanation for why first year and fourth year respondents were selected. Why, and what are possible implications, for instance for the statical analysis as well as the findings? Whether there was any difference between the groups/levels is a question that comes to mind. In my opinion this needs to be better addressed in relation to the validity of the findings. Response 2: Thank you for your comments. As we had convenient access to year 1 and year 4 students, these were the participants that were originally recruited for the study. As learners in the university often take courses with their own classmates and mingle usually with students in their own grade level, we ended up with only year 1 and year 4 students as participants. As we were given a small funding to recruit 40 students, we opened the recruitment to these two years of students as they likely may have lower course loading than year 2 and year 3 students.    Comment 3: By the way, is it Pearson's or Spearman that is used in the correlation analysis? Would not the latter be most appropriate? Response 3: Thank you for your comments. We used Spearman correlation in the analysis process.  The data for CGPA, CKT scores, UVLT scores, LVLT scores, and the AWL scores were not normally distributed as shown by the results of Shapiro-Wilk tests: CGPA (p = .018), CKT scores (p = .008), UVLT scores (p < .001), LVLT scores (p = .004), AWL scores (p = .005). Therefore, two-tailed non-parametric Spearman’s rho correlations were conducted to investigate the relationship among CGPA, CKT scores, and receptive aural/written vocabulary scores (Larson-Hall, 2010). The relevant information can be found on page 8, section 3.4 Data Analysis, line 821-828.   Comment 4: I think that the CEFR construct with its focus on reading, writing, listening and speaking could be explained a bit, it should come as no surprise that it is not always a consistent predictor of academic skills. Indeed, even the TOEFL and IELTS tests are not either. This makes the finding, the utility of supplement testing with appropriate vocabulary tests particularly compelling Response 4: Thank you for your insightful comment. We appreciate your suggestion to elaborate on the CEFR construct with its focus on reading, writing, listening, and speaking. We agree that these skills, while central to language proficiency, may not consistently predict academic skills across all contexts. This variability underscores the importance of our finding: supplementing traditional proficiency tests with specific vocabulary assessments can provide a more comprehensive evaluation of a student's readiness for academic success in an EMI context. In response to your feedback, we included a brief explanation of the CEFR framework in our revised manuscript to clarify its components and their potential limitations in predicting academic skills. We added some sentences in the text as follows: “Although the CEFR provides general guidelines and a set of descriptors with its focus on reading, writing, listening, and speaking, it cannot refer to any specific language or context of use (Piccardo, 2020). Therefore, this makes the utility of supplement testing with appropriate vocabulary tests (i.e., UVLT, LVLT) particularly compelling. Specifically, this variability underscores the importance of our study: supplementing traditional proficiency tests with specific vocabulary assessments can provide a more comprehensive evaluation of a student's readiness for academic success in an EMI context.” The relevant information can be found on page 4, section 2.2 Mapping Vocabulary Levels to the CEFR, line 441-448.

Reviewer 2 Report

Comments and Suggestions for Authors

See my attached review. I enjoyed reading your paper. Keep up the good work!

Comments for author File: Comments.pdf

Author Response

Response to Reviewer 2 Comments
1. Summary
Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions in track changes in the re-submitted files.
2. Point-by-point response to General Comments and Suggestions for Authors
Comment 1: The structure of the paper’s introductory sections is, in my opinion, not easy to follow. The research questions are presented directly after the introduction. They are followed by a literature review that to some extent seems to repeat information from the introduction, and also in some places refers to the present study which is, of course, not part of any previously published literature. I think the paper would be much more readable if all context, introduction, and previous research comes first, before the research questions are presented as emanating from gaps of knowledge in previous research. I suggest using the same structure as in the article by Reynolds et al. (2022), cited by the authors, i.e., an introduction/background followed by a problem statement and research questions appearing immediately before the methods section.
Response 1: Thank you for pointing this out. We agree with your comment and have made the necessary adjustments. As suggested, we restructured the Introduction and the Literature Review. This included adding a Problem Statement and Research Questions section to the Literature Review, as well as incorporating topic and concluding sentences in several paragraphs. Specifically, the research questions are now presented at the end of the Literature Review section, immediately followed by the Methodology section. This change can be found on page 3, and page 5, section 2.3 Problem Statement and Research Questions. Thank you for pointing this out.
Comment 2: Although VLT seems to have been designed primarily for diagnostic purposes, indicating at what level a student should focus their vocabulary learning, overall scores on the VLT may still be useful in correlational research. For example, McLean, Kramer and Beglar (2015) used Rasch person ability estimates based on all LVLT items (i.e., an overall score) when validating their test against the TOEIC. I suggest that the authors report both overall scores with their distributions, as well as level mastery. Correlations between vocabulary and CEFR should preferably be done using total scores and not level mastery, the latter being useful mainly for guiding teachers and students. Another reason why total score correlations are valuable is that they may be included in future meta-analyses seeking to establish associations between CEFR and its linguistic correlates. For this reason (and others), the authors should also report reliability estimates for the vocabulary scores, for example coefficient alpha. Response 2: Thank you for pointing this out. We agree with this comment. First of all, we have included the overall scores with their distributions in the passage. This change can be found on page 9, Table 4. Moreover, the correlation between vocabulary and the CEFR was done using total scores of the UVLT, the LVLT and the CKT. Finally, coefficient alpha for the UVLT and the LVLT were obtained, which indicated a fairly high level of internal consistency for vocabulary scores with this specific sample. The coefficients were reported on page 8, Section 3.4 Data Analysis, line 828-832.
Comment 3: I am not sure that this fits into the scope of the study, but I would find it interesting to see a regression model exploring how much better a combination of aural and written vocabulary would predict CEFR-level, than any single vocabulary measure. Admittedly, the sample is small (as the authors point out, line 526) but with only two predictors (the vocabulary tests) there would be 20 cases per independent variable, considered enough in many statistics textbooks. Response 3: Thank you for pointing this out. We agree with this comment. Therefore, we have included regression analyses in the revised manuscript. We described the procedure of conducting the regression analysis in section 3.4 Data Analysis. This change can be found on page 7-8, line 784-840. The results of the regression analyses are in section 4.5 Regression Analyses. This change can be found on Table 7 and the interpretation of the results are on page 11, line 1449-1457. Additionally, we removed one limitation in section 7 Limitation and Future Research. This change can be found on page 14, line 1977.   Comment 4: I was disappointed by the very sloppy referencing. For example, in the discussion on page 10, twelve of the cited works are missing in the reference list. There are also several (I stopped counting at seven) posts in the reference list that do not appear in the main text. It is understandable if a few references are missing in a manuscript. In this case, however, the omissions are so many that it seems like outright negligence. May I suggest that, for future work, the authors familiarize with referencing software (e.g., Zotero is a very useful, free tool) to ensure that their references are always up to date. Response 4: Thank you for pointing this out. We have gone through the writing again and added the previously missing references to the reference list.

 

3. Point-by-point response to Specific Comments and Suggestions for Authors

Comment 1: 32. Reynolds and Yu (2018) is missing in the reference list.

Response 1: Thank you for pointing this out. We have added the reference to the reference list. This change can be found on page 18, reference no.62.

 

Comment 2: 34. Reynolds et al. (2023) is missing in the reference list.

Response 2: Thank you for pointing this out. We have added the reference to the reference list. This change can be found on page 18, reference no.63.

 

Comment 3: 76. I find it odd to think of CEFR-level varying in relation to vocabulary level. It sounds as if the variable vocabulary level can be manipulated like in an experiment, which is clearly not the case here. I would probably have asked something like “To what extent does the English proficiency of EMI students reflect their mastery of receptive aural vocabulary levels?” Or perhaps, “… to what extent is English proficiency associated with mastery …”. 

Response 3: We agree with this comment. Therefore, we have revised the corresponding sentences as follows: “RQ1: To what extent is the English proficiency of EMI students (i.e., CEFR levels) associated with their mastery of receptive aural vocabulary levels?” and “RQ3: To what extent is the English proficiency of EMI students (i.e., CEFR levels) associated with their mastery of receptive written vocabulary levels?” These changes can be found on page 5, section 2.3 Problem Statement and Research Questions, line 561-562 and line 565-566.

 

Comment 4: 95. The authors seem to refer to their own study here, but the rubric on line 84 says that this is a literature review. The entanglement of their own study and the background literature can be avoided by restructuring the introduction/background/aims, see my general comment above on this issue.  

Response 4: Thank you for pointing this out. We agree with this comment. As suggested, we restructured the Introduction and the Literature Review. This change includes moving the research gap and research questions that were previously in the introduction section to the new-added Problem Statement and Research Questions, section in the Literature Review. Please find the report of changes in Response 1 in the Response to General Comment section.

 

Comment 5: 96-98. I find this sentence confusing.

Response 5: Thank you for pointing this out. We have revised the sentence as follows: “knowing their vocabulary levels mastery allows for the comparison between students’ aural/written vocabulary levels mastery and the frequency levels of words required to comprehend EMI lectures and instructional materials of different disciplines. With this information at hand, key stakeholders can more confidently predict the students’ readiness for learning in specific EMI programs.” For the coherence of the text, we have moved this part to the second paragraph of section 2.1 Vocabulary Mastery in EMI Programmes. This change can be found on page 3, line 244-249.

 

Comment 6: 119. “required for satisfactory listening comprehension”, however, this paragraph seems to be about reading comprehension (cf. line 114). The following paragraph (beginning on line 126) seems to be about listening. Please rearrange so that each paragraph deals with one topic.

Response 6: We agree with this comment. We have moved the relevant sentence to the paragraph concerning listening comprehension. The changes can be found on page 3, line 255-258.

 

Comment 7: 128. Clarify what “percentage of variance” refers to. Is it percentage of variance in listening comprehension outcomes explained by vocabulary knowledge?

Response 7: Thank you for pointing this out. Yes, it means the percentage of variance in listening comprehension outcomes explained by vocabulary knowledge. The sentence was revised as the following: “However, the explanatory power of vocabulary knowledge in the success of L2 listening comprehension varies from 13% (Wang and Treffers-Daller, 2017) to 59% (Cai, 2020) among previous studies examining the correlations between L2 vocabulary and listening (Wang, 2023).” This change can be found on page 3, line 258-261.

 

Comment 8: 129. Wang and Treffers-Daller (2017) is missing in the reference list.

Response 8: Thank you for pointing this out. We have added the reference to the reference list. This change can be found on page 19, reference no. 82.

 

Comment 9: 136. All three works cited here are missing in the reference list.

Response 9: Thank you for pointing this out. We have added the references to the reference list. These changes can be found on page 17, reference no.5 (Bonk, 2000); page 19, reference no. 71 (Staehr, 2008). The reference of Wang and Treffers-Daller (2017) has been addressed in comment 8.

 

Comment 10: 147. “contributed to the 59% of variance” sounds odd.

Response 10: Thank you for pointing this out. We have revised the sentence as the following: “Moreover, Cai (2020) investigated the relationship between L2 learners’ receptive aural vocabulary knowledge and L2 listening comprehension and suggested that L2 learners’ receptive aural vocabulary knowledge contributed to the prediction of 59% of the variance in their L2 listening comprehension.” This change can be found on page 3, line 259-263.

 

Comment 11: 148. Both cited works are missing in the reference list.

Response 11: Thank you for pointing this out. We have added the reference to the reference list. These changes can be found on page 18, reference no.45 (McLean et al., 2015); page 19, reference no. 80 (Wallace, 2020).

 

Comment 12: 152. Du et al. (2020) is missing in the reference list.  

Response 12: Thank you for pointing this out. It should be Du et al. (2022) and the reference was already included in the reference list. The change of year can be found on page 3, line 284.

 

Comment 13: 165. British Council (2021) – is this the same as British Council (n.d.) in the reference list?

Response 13: Thank you for pointing this out. It should be British Council (n.d.) and the reference was already included in the reference list. The change of year can be found on page 4, line 456.

 

Comment 14: 179. Nation (2024) should be Nation (p.c.) if I interpret the footnote correctly.

Response 14: Thank you for pointing this out. “Nation (2024)” has been corrected to “Nation (p.c.)”. These changes can be found on page 4, line 470, and Table 1.

 

Comment 15: 184. In Table 1, are the reported figures lemmas or word families?

Response 15: Thank you for your comment. For X-Lex score (Meara and Milton, 2003) and X-Lex score (adapted from Milton, 2009), the reported figures should be lemmas. For the suggested vocabulary size from Nation, the reported figures should be word families. Please find these changes on page 4, Table 1.

 

Comment 16: 219. It is customary to report reliability coefficients if feasible. It should not be complicated to compute coefficient alpha for the vocabulary tests. The authors may want to check score unidimensionality (an assumption for coefficient alpha) or refer to literature that has investigated dimensionality of these tests. I think McLean et al (2015) reported that LVLT scores were strongly unidimensional. 

Response 16: Thank you for pointing this out. We have reported the dimensionality of the two tests by referring to literature. The changes can be found on page 6, line 627 (dimensionality of the LVLT); and line 653-654 (dimensionality of the UVLT).

 

Comment 17: 220. I would suggest presenting the instruments in the same order as they occur in the results section, i.e., LVLT first and then UVLT.

Response 17: We agree with this comment. Therefore, we have changed the order as suggested: the LVLT was presented before the UVLT. This change can be found on page 6, line 424-663.

 

Comment 18: 228. Could the authors briefly explain the rationale behind using 29 out of 30 as the mastery criterion for the first three 1000-word? If I remember correctly, most other VLT-studies have used a lower criterion score. Given that there is always measurement error, I think 29/30 seems a bit harsh. 

Response 18: Thank you for your comments. The mastery criterion of 29/30 was suggested by the developers of the UVLT (Webb et al., 2017, p. 56). It was claimed that the first 3,000 word families account for a very large percentage of English and thus having a great value in making further lexical and language development. Therefore, a strict threshold was employed in this study to ensure that learners had achieved “a near perfect knowledge of their form-meaning connections” (Webb et al., 2017, p. 56) before moving on to the next level (i.e., 2,000 and 3,000 level).

 

Comment 19: 234. Both references are missing in the reference list.

Response 19: Thank you for pointing this out. We have added the reference to the reference list. This change can be found on page 18, reference no.34. The reference of McLean et al (2015) has been added based on comment 11.

 

Comment 20: 237. “measuring 1,000-word frequency on one level” sounds unclear to me.

Response 20: Thank you for pointing this out. We have revised the sentence as follows: “The first five sections correspond to the five frequency levels of the most frequent 5,000 word families (e.g., section one measures the most frequent 1,000 word families, section two measures the most frequent 2,000 word families), and each of these five sections contains 24 items.” This change can be found on page 6, paragraph 1, line 628-632.

 

Comment 21: 248. Were the same criterion score used with LVLT as with UVLT? Please add.

Response 21: Thank you for pointing this out. The criterion used with LVLT was in accordance with Schmitt et al.’s (2001) 26/30 criterion per level. The following sentences were added to 3.2.2 Updated Vocabulary Levels Test for clarification: “To determine participants’ mastery level of aural vocabulary, a 21/24 cut-off score was used for the first 5,000 frequency levels and a 26/30 cut-off score was used for mastery of the AWL in accordance with Schmitt et al.’s (2001) suggested 26/30 criterion per level. According to Schmitt et al. (2001), learners were presumed to have mastered the higher frequency levels if they had mastered the lower frequency levels and this guideline was also followed in this study. A participant’s aural vocabulary mastery level was deter-mined by the highest level that a participant reached mastery.” This change can be found on page 6, line 642-649.

 

Comment 22: 281. I would appreciate a table with ranges, means, standard deviations, and skewness for both vocabulary tests. For the UVLT, mean and sd are reported in passing on line 358 but it would be better to present this information together with LVLT descriptives. 

Response 22: We agree with this comment. Therefore, we have added a Table 4 showing the ranges, means, standard deviations, and skewness for both UVLT and LVLT tests. Please find the changes on page 9, Table 4.

 

Comment 23: 307. The column headings in Table 4 look odd. Shouldn’t they be 1k, 2k, 3k, 4k and 5k? I suggest that the authors add a column with the total number of participants at each CEFR-level.  

Response 23: We agree with this comment. Therefore, we have added the column with 4k frequency level and columns with the total number of participants at each CEFR-level for first 5k levels of LVLT and AWL. These changes can be found on page 8 and 9, Table 3.

 

Comment 24: 321. Did the authors explain why they include CGPA in this study? Neither the title of the manuscript, nor the research questions, suggest that this paper is about the relationship between CGPA and other variables. 

Response 24: Thank you for your comment. We included CGPA in this study because we would like to see whether the language tests aligned with the CEFR levels can be used as a tool for informing if L2 learners’ vocabulary knowledge is sufficient for their success in a specific learning environment. Therefore, we used CGPA as a criterion to directly measure learners’ academic success.

 

Comment 25: 339. I would suggest having just one table of correlations that includes both UVLT and LVLT. Now, Table 7 repeats some information from Table 5 (i.e., correlations between CEFR and CGPA). A common table of correlations would also allow readers to see the association between aural and written vocabulary. Also, I am not sure, but my impression is that the reported correlations with CEFR were computed using level mastery (i.e., an ordinal, six-point scale), rather than overall vocabulary scores (i.e., a continuous variable). Artificially coding a continuous variable into just six categories means discarding a lot of valuable information in the scores. 

Response 25: We agree with this comment. Therefore, we have combined Table 5 and Table 7 into one (i.e., Table 5). Moreover, the correlations with the CEFR are now computed using the overall vocabulary scores of the UVLT and LVLT tests. The combined table can be found on page 10, Table 5.

 

Comment 26: 376. The correlation coefficient mentioned here between CEFR and UVLT (.602) does not appear in Table 7 where it says .559 instead. 

Response 26: Thank you for your comment. The correlation coefficient has changed to .682 after using the overall scores of the UVLT and the CEFR in the correlation analysis. This change can be found on page 10, Table 5.

 

Comment 27: 389. Many (or most) references on this page cannot be found in the reference list.

Response 27: Thank you for pointing this out. We have added the references (Dang, 2022; Lin and Morrison, 2010; Qian and Lin, 2019; Rafique et al., 2023; Zareva et al., 2005; Laufer, 2013; Matthews and Cheng, 2015; Masrai and Milton, 2018) to the reference list. These changes can be found on page 17: reference no.15 (Dang, 2022); page 18: reference no. 37 (Laufer, 2013); reference no.39 (Lin and Morrison, 2010); reference no.42 (Masrai and Milton, 2018); reference no.43 (Matthews and Cheng, 2015); reference no.60 (Qian and Lin, 2019); reference no.61 (Rafique et al., 2023); page 19, reference no.90 (Zareva et al., 2005).

 

Comment 28: 434. I am not following here. The correlation between aural vocabulary and CGPA was statistically significant, or?

Response 28: Thank you for your comment. The idea of the sentence “The statistically insignificant correlation between L2 learners’ receptive aural academic vocabulary mastery and their CGPA suggested that the role of receptive aural academic vocabulary mastery might not as significant as that of written academic vocabulary knowledge in tertiary education achievement.” is that the correlation between aural academic vocabulary and CGPA was not statistically significant. While written academic vocabulary was found to be positively correlated to tertiary education achievement (Masrai and Milton, 2018), the correlation between aural academic vocabulary and CGPA was statistically insignificant. Therefore, aural academic vocabulary knowledge might not play a significant role as written academicvocabulary in tertiary education achievement.

 

4. Additional clarifications

After conducting the correlation analyses using the test scores of the LVLT, the AWL, the UVLT, and the CKT, we found that the correlation between the AWL scores and CGPA, and the correlation between the UVLT scores and CGPA are statistically significant. Therefore, we made some adjustments in the section 4.2 (this change can be found on page 10, line 1041-1059), section 4.4 (this change can be found on page 11, line 1146-1153), and section 5 Discussion (this change can be found on page 12-13, line 1687-1842).

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript has been much improved and the authors have dealt with my comments on the previous version in a satisfactory manner. The manuscript should be ready for publication, but I spotted a few minor issues:

line 290: "Table 2. The correlation between ..." I would prefer to to call it a "comparison" or "relationship" of EnglishScore and the other tests, because "correlation" is used elsewhere in this paper in its statistical meaning (when correlating variables in the study).

line 349: Why are there six levels of LVLT (0k, 1k, 2k, 3k, 4k and 5k)? I don't understand what the 0k level refers to. There should only be five levels, plus the AWL. It is strange that the authors present data for a "0k"-level because such a level does not exist in the test. The LVLT has five levels (1k, 2k, 3k, 4k, and 5k) plus the AWL. In Table 4, with descriptive statistics, the levels are correctly reported.

Line 419: Same as in my previous comment: There are only five levels in the UVLT (1k, 2k, 3k, 4k, and 5k). Where does the 0k-level come from? This issue must be resolved before publication.

It was interesting to see the correlations between the vocabulary measures and the CKT, as well as the regression models. A large part of the variance in communicative proficiency can apparently be explained by receptive vocabulary knowledge. This is of course well known before, but I like how this study included both written and auditory vocabulary. It is interesting that the two modalities were so closely correlated. I agree with the authors' conclusion that EMI universities could benefit from using (quick and cheap) vocabulary tests along with their standard procedures. The reliability coefficients of the vocabulary tests in this study were on the low side, but that may be explained by the truncated sample (the participants were already selected for university) with a rather marked negative skew (the tests were relatively easy for the students). If a sample had been obtained by applicants to the university (before being admitted), there may have been a larger variation in vocabulary scores which should increase reliability.

Author Response

Comments and Suggestions for Authors

The manuscript has been much improved and the authors have dealt with my comments on the previous version in a satisfactory manner. The manuscript should be ready for publication, but I spotted a few minor issues:

line 290: "Table 2. The correlation between ..." I would prefer to to call it a "comparison" or "relationship" of EnglishScore and the other tests, because "correlation" is used elsewhere in this paper in its statistical meaning (when correlating variables in the study).

Response 1: Thank you for pointing this out. We agree with this comment. Therefore, we changed the title of Table 2 into “The relationship between ...” This change can be found on page 7, Table 2, line 772.

line 349: Why are there six levels of LVLT (0k, 1k, 2k, 3k, 4k and 5k)? I don't understand what the 0k level refers to. There should only be five levels, plus the AWL. It is strange that the authors present data for a "0k"-level because such a level does not exist in the test. The LVLT has five levels (1k, 2k, 3k, 4k, and 5k) plus the AWL. In Table 4, with descriptive statistics, the levels are correctly reported.

Response 2: Thank you for pointing this out. We agree with this comment. Therefore, we removed the 0k column in Table 3. Additionally, we added notes below the table, indicating the number of participants that did not master any of the five frequency levels of the LVLT. This change can be found on page 8, Table 3, line 847-850.

Line 419: Same as in my previous comment: There are only five levels in the UVLT (1k, 2k, 3k, 4k, and 5k). Where does the 0k-level come from? This issue must be resolved before publication.

Response 3: Thank you for pointing this out. We agree with this comment. Therefore, we removed the 0k column in Table 6. Additionally, we added notes below the table, indicating the number of participants that did not master any of the five frequency levels of the UVLT. This change can be found on page 10-11, Table 6, line 1061-1398.

It was interesting to see the correlations between the vocabulary measures and the CKT, as well as the regression models. A large part of the variance in communicative proficiency can apparently be explained by receptive vocabulary knowledge. This is of course well known before, but I like how this study included both written and auditory vocabulary. It is interesting that the two modalities were so closely correlated. I agree with the authors' conclusion that EMI universities could benefit from using (quick and cheap) vocabulary tests along with their standard procedures. The reliability coefficients of the vocabulary tests in this study were on the low side, but that may be explained by the truncated sample (the participants were already selected for university) with a rather marked negative skew (the tests were relatively easy for the students). If a sample had been obtained by applicants to the university (before being admitted), there may have been a larger variation in vocabulary scores which should increase reliability.