The Online Effects of Processing Instruction on the Acquisition of the English Passive Structure

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Review of 

The online effects of processing instruction on the acquisition of the English passive structure

This study investigated the effects of two types of instruction on Persian L1-English L2 learners’ interpretation of English passives. Using a self-paced reading experiment, the study’s results showed greater accuracy in interpretation in the groups receiving instruction compared to the control group. No difference was found between groups with respect to reaction times, however, the difference in accuracy between groups persisted for three weeks. The results are interpreted as supporting the efficacy of providing learners with structured input to change their processing strategies.

The paper is well-written, and the methods and results especially are clearly explained. My main concerns relate to the methods and the discussion, but a few comments pertain to the abstract and background as well. These should be addressed. A few typos and possible re-phrasings are identified below.

Main comments:

Abstract: PI should be defined in the abstract

Background: p. 6, l. 278: Some speculation about what might have caused the different results in Lee et al 2020 vs Malorvrh et al. 2020 would be interesting here – e.g. were L1s different and in which ways?

Methods:

8. 8. It’s not clear to me how the instructional treatment differed from simply additional exposure to the structure. That is, I worry that there is a risk that the difference between groups is merely a result of additional exposure to passive constructions rather than due to the instructional treatment. A rationale should be provided for not offering a corresponding amount of exposure to passives (but without instruction) to the control group. This issue should also be included in the discussion.
9. l. 341: What exactly were participants in the PI group told about the processing problem?
10. 8, l. 344. “No explicit feedback was provided to participants” – I don’t understand this, because they were getting correct/incorrect feedback on their picture selection (out of 2 pictures). What other feedback could be provided?

Overall, it’s not clear to me how the Instructional Treatment applied in the study relates to the discussion of what processing instruction is (on p. 2). On p. 2, several steps seem to be involved. Please clarify how the treatment was operationalized in the study and why.

I don’t understand why the reaction time measure is from onset of picture presentation to picture selection if the issue of interest is online processing. Presumably, participants are doing part of the processing as there are reading the sentence (as is briefly mentioned in the discussion). An entire body of literature exists on online processing (including self-paced reading studies) of L2 vs. L1 learners. It seems that a more thorough discussion of this literature is warranted, and a justification of the study’s reaction time measure should be provided. Or, better yet, if you have the data, why not analyze the word-by-word reaction times instead?

Discussion:

Other than Input Processing Theory, what evidence is there that learners (here and elsewhere) in fact use the wrong processing strategy for passives? Is there any evidence that learners are not using the same strategy as L1 users, just slower?

The results of the present study suggest no difference between PI/SI groups. Has anyone tested what happens if participants only receive instruction on how to move away form lexico-semantic processing and are not given structured input? Understanding this would help clarify whether the lack of difference between SI/PI groups here is simply a ceiling effect (i.e. there is no additional benefit to PI on top of SI, but there might be a benefit to PI over nothing).

The modality question seems to come out of nowhere in the discussion, and its relevance is not entirely clear to me. This should be better motivated throughout the paper, and the relevance for understanding the current results should be made clear.

Typos, etc.

1. 1, l. 42: “ This, more recently, research on PI has witnessed” > should this say “this more recent research on PI…”?

p.2, l. 50: “grounded on” > should this be “grounded in”?

2. 2., l. 78-79 : “aims at assisting learners to make correct form-meaning connections” > should this be “aims to assist learners in making ….”?
3. 6, l. 245 “affectively” > should this be “effectively”?
4. 6, l 262, and 276: “enquiry” > “inquiry”?

Author Response

• Comment: PI should be defined in the abstract

Response: Comment accommodated!

• Comment: Background: p. 6, l. 278: Some speculation about what might have caused the different results in Lee et al 2020 vs Malorvrh et al. 2020 would be interesting here – e.g. were L1s different and in which ways?

Response: We could not find any potential reasons for why the two studies found different results, as all conditions, including participants’ features, were identical in both studies, and the authors did not explain the contrary findings in the studies. This actually further motivated us to investigate the long-term effects of PI.

• Comment: It’s not clear to me how the instructional treatment differed from simply additional exposure to the structure. That is, I worry that there is a risk that the difference between groups is merely a result of additional exposure to passive constructions rather than due to the instructional treatment. A rationale should be provided for not offering a corresponding amount of exposure to passives (but without instruction) to the control group. This issue should also be included in the discussion.

Response: Exposure to the passive structure (although without instruction) would indeed count as a type of instruction (input flood). Furthermore, previous research has indicated that mere exposure to the target structure does not result in any significant achievements in sentence interpretation (e.g., ChiuChiù & Benati, 2020; Issa & Morgan-Short, 2019). Therefore, following similar studies (e.g., Benati, 2023; Benati & Batziou, 2017; Benati & Chan, 2023; Cadierno, 1995; Cheng, 2004; Lee, 2015; VanPatten & Wong, 2004), a decision was made that the control would only take tests.

• Comment: What exactly were participants in the PI group told about the processing problem?

Response: Comment accommodated! “More specifically, the participants in the PI group were informed about the potential incorrect interpretation of the first noun or pronoun as the agent (doer) of the action rather than the receiver of the action in passive sentences.”

• Comment: “No explicit feedback was provided to participants” – I don’t understand this, because they were getting correct/incorrect feedback on their picture selection (out of 2 pictures). What other feedback could be provided?

Response: The participants could, for example, get feedback in the form of metalinguistic explanations about why their answer was correct or incorrect. However, they did not.

• Comment: Overall, it’s not clear to me how the Instructional Treatment applied in the study relates to the discussion of what processing instruction is (on p. 2). On p. 2, several steps seem to be involved. Please clarify how the treatment was operationalized in the study and why.

Response: As explained in section 3.3, the instructional treatment encompassed SI activities (72 tokens) for the SI group and SI activities + explicit information about the processing problem for the PI group. As mentioned in this section, the SI activities were the same for both groups. Therefore, the only difference between the two instructional treatments would be the availability of explicit information for the PI group, but not for the SI group. For more clarity, we have now added what specifically the explicit information provided for the PI group was. We have also added that SI activities were designed following the guidelines provided by Lee and VanPatten (2003).

• Comment: I don’t understand why the reaction time measure is from onset of picture presentation to picture selection if the issue of interest is online processing. Presumably, participants are doing part of the processing as they are reading the sentence (as is briefly mentioned in the discussion). An entire body of literature exists on online processing (including self-paced reading studies) of L2 vs. L1 learners. It seems that a more thorough discussion of this literature is warranted, and a justification of the study’s reaction time measure should be provided. Or, better yet, if you have the data, why not analyze the word-by-word reaction times instead?

Response: We agree that the word-by-word reaction time would be a better indicator of online input processing. However, this study is part of a larger research project, and the data presented in this study is the preliminary analysis of the data collected. We are planning to present the word-by-word reaction times along with some other variables in our next paper. In the Conclusion section, we have actually acknowledged that the word-by-word reaction time (reading time) would provide a better picture of online input processing.

• Comment: Other than Input Processing Theory, what evidence is there that learners (here and elsewhere) in fact use the wrong processing strategy for passives? Is there any evidence that learners are not using the same strategy as L1 users, just slower?

Response: The Input Processing Theory is the main theory explaining how L2 learners process input. There is also empirical evidence showing that L2 learners use the wrong processing strategy for passives as represented in their eye-movement patterns (e.g., Benati, 2020, 2021; 2022). Additionally, there is evidence suggesting that L2 learners do not process passives similar to native speakers (e.g., Lee & Doherty, 2019). Therefore, it is not only being slower in processing passives, but also processing the passive structure (among other structures) inaccurately.

• Comment: The results of the present study suggest no difference between PI/SI groups. Has anyone tested what happens if participants only receive instruction on how to move away form lexico-semantic processing and are not given structured input? Understanding this would help clarify whether the lack of difference between SI/PI groups here is simply a ceiling effect (i.e. there is no additional benefit to PI on top of SI, but there might be a benefit to PI over nothing).

Response: There are a number of studies showing that only receiving instruction on how to move away from the default less-than-optimal processing strategies does not result in any improvements in processing the structure (e.g., Benati 2004a, 2004b; VanPatten & Oikkenon, 1996; Wong, 2004). This has been briefly mentioned in Sections 2.2 and 2.3. In fact, the results of our study also support this. The only difference between PI and SI was the provision of explicit information about the processing problem to the PI group, but not the SI group. Therefore, the lack of a significant difference between PI and SI shows no additional benefits for PI over SI, which we have also mentioned in the discussion section.

• Comment: The modality question seems to come out of nowhere in the discussion, and its relevance is not entirely clear to me. This should be better motivated throughout the paper, and the relevance for understanding the current results should be made clear.

Response: Input modality in testing is actually not part of the discussion. We have mentioned it as a suggestion for future research. As we have mentioned at the end of the article, there is evidence suggesting that processing auditory input is generally more challenging than processing written input. However, there is dearth of research in PI literature addressing the role of input modality (i.e., auditory vs. written input) in processing the input. This question is particularly relevant to PI research as PI is basically about making learners better input processors. Thus, we believe future research should explore whether and the extent to which learners’ processing of input would depend on the modality of the input. In other words, future research should investigate whether learners receiving PI would perform differently in processing input when the input is provided in written or auditory mode.

• Typos

All corrected!

Reviewer 2 Report

Comments and Suggestions for Authors

Please see attached review.

Comments for author File: Comments.pdf

Author Response

• Comment: Define PI and use the term “processing instruction” in the abstract, before using the acronym PI. Define “structured input” briefly in abstract (a phrase would be sufficient).

Response: Comment accommodated!

• Comment: Overall I think the introduction needs another revision where the authors take care to use formal language in their prose. The whole introduction needs to be more academic sounding.

Response: Comment accommodated! We have made revisions in the introduction.

• Comment: A good definition of PI needs to start the introduction. I know you explain it well later in the introduction, but a brief clause or sentence in the introduction would be helpful. An example of PI would also be helpful here.

Response: We agree that, as an alternative, the introduction could start with a definition of PI. However, based on how we have structured the introduction, starting it with a definition of PI would disrupt coherence/cohesion. We have, however, now added a brief description of PI at the end of the first paragraph.

Since examples are already provided in Section 2.2, which follows shortly after the Introduction, we believe including them in the Introduction as well would be redundant.

• Comment: Some of the sentences in the introduction are excessively wordy. An example is here on lines 37 – 41. Consider changing to “In other words, learners’ reaction times to the stimuli reveals the level of cognitive effort involved in processing the input. Since the stimuli in these tests are brief and focus on comprehension, learners would not need to rely on their explicit knowledge.”

Response: Comment accommodated!

• Comment: Page 1, paragraph 2: Can you explain very briefly why learners don’t rely on explicit knowledge during these tasks?

Response: Thanks for the comment. We have now revised our statement to make it more accurate. Indeed, assuming that learners would definitely rely on their implicit knowledge would be incorrect. We have now acknowledged that learners may still rely on their explicit knowledge and have provided further explanations.

• Comment: Page 2, top paragraph (lines 43-47)—Change “shift of attention” to “shift of focus”

Response: Comment accommodated!

• Comment: Background

 In this first paragraph, explicitly use the name of Van Patten’s model (Input Processing theory). Also provide a brief definition of intake and input.

Response: Comment accommodated!

• Comment: 2. It’s not completely clear if you are quoting previous writing when you talk about PI features (you have the page number and quotes, but I’m not sure if this is a direct quote or not. Either way, I think this section would be better stated in your own words, not quoting so much of a previous study. It would also be helpful if you gave some examples as you talked about each of the features. The same is true for the next page (Page 3) where you quote about Structured input.

Response: These are actually direct quotes. We have not paraphrased the quotes as we are referring to the key features of PI and guidelines for designing SI, as proposed by the authors. We believe paraphrasing would not reflect all the features fully.

Giving examples for each feature individually is not really possible. However, as you have mentioned in another comment, we have provided examples for SI activities after talking about features. Moreover, we have now shared the instructional materials in the supplementary materials section. Readers interested in a full picture of SI activities, can refer to the supplementary materials.

• Comment: I like that you gave an example of what structured input looks like. Can you explain why they only receive a “yes” or “no” feedback and why this is part of Information Processing Theory.

Response: The reason feedback is limited to only “correct/incorrect” or “Yes/No” is because there should be no explicit information provided to the learners in SI activities. As explained in section 2.2, the only explicit information learners may receive is for the PI group, which comes before receiving SI practice. Therefore, feedback is provided only in the form of “correct/incorrect” or “Yes/No”. This is indeed what has been done throughout the literature. Any other type of feedback provided would be counted as another variable to be investigated as in Sanz (2004) and Sanz & Morgan-Short (2004).

• Comment: Page 4: Change “in production equal to the DG group” to “in production was equal to the DG group”. For more clarity, however, we have now revised the sentence.

Response: Changing “in production equal to the DG group” to “in production was equal to the DG group” would make the sentence grammatically incorrect.

• Comment: 5: Can you explain better how online processing “may also tap into learner’s implicit knowledge”?

Response: We have now revised the sentence to “may also provide a picture of the development of learners’ implicit knowledge”. Following one of your previous comments, we have now explained this further in Section 1 (Introduction).

• Comment: Page 6: change “suggesting that PI was affective in influencing” to “suggesting that PI was effective in influencing”

Response: Comment accommodated!

• Comment: Page 6: Can you explain very briefly what “guided-inductive instruction” is?

Response: Comment accommodated!

• Comment: Throughout the introduction and background, there needs to be a better cohesion of why cognitively PI and SI work and what the difference is between them. I wonder about how you defined “PI” and how you used it in instruction for this study—has this method of using PI in instruction been used before?

Response: As explained in section 2.2 (and now further elaborated on), SI is a subset of PI. That is PI includes two components: 1) explicit information about the target structure and the processing problem and 2) SI practice. In sections 2.2 and 2.3, we have explained that PI assists learners to make accurate form-meaning connections during processing input for comprehension, through pushing them away from their less-than-optimal input processing strategies. As predicted by the Input processing Theory, L2 learners rely on their default input processing strategies to comprehend input. These strategies, however, are not optimal and may result in inaccurate intake. PI, on the other hand, pushes learners away from their default non-optimal input processing behaviour. This is done, mainly through providing SI activities.

As explained in section 3.3, in this study, PI included explicit information about the target structure followed by SI activities. This is actually how PI has been used throughout the literature, with the only difference that in some studies (but not all), explicit information provided to the PI group included information about both the target structure and the processing problem.

• Comment: Can you tell us more about the participants—ages mean and standard deviation, levels of proficiency, gender, etc?

Response: Comment accommodated!

• Comment: Do you know how many or what words the learners had difficulty remembering? Can you let us know the data so we can see what words you used?

Response: Overall, 17% of the participants had difficulty remembering 3-6% of the words. Materials have now been shared in the supplementary section. However, please kindly note that pictures have been removed for copyright compliance.

• Comment: I’m wondering why you didn’t have the control group do any kind of instruction?

Response: Providing instruction to the control group would count as a type of instructional treatment, which would not make it Control then. Even mere exposure to the passive structure (although without instruction) would indeed count as a type of instructional treatment (input flood). Furthermore, previous research has indicated that mere exposure to the target structure does not result in any significant achievements in sentence interpretation (e.g., ChiuChiù & Benati, 2020; Issa & Morgan-Short, 2019). Therefore, following similar studies (e.g., Benati, 2023; Benati & Batziou, 2017; Benati & Chan, 2023; Cadierno, 1995; Cheng, 2004; Lee, 2015; VanPatten & Wong, 2004), a decision was made that the control would only take tests.

• Comment: Was the Anova run a repeated measures anova?

Response: The ANOVA run was a mixed-design ANOVA, followed by pairwise comparisons with Bonferroni correction.

• Comment: The tables and figures are very helpful for understanding the data. On Table 4, however, the last row (significance) needs to be revised.

Response: Table 4 revised!

• Comment: Can you refer back to your research questions in the discussion and talk about what the answers to these questions was? You do this implicitly, but can you explicitly tell us what you found for each question?

Response: We agree that, overall, explicitly referring to the research questions would be helpful. However, it should be noted that, in this study, each research question focuses on the effects of either PI (Q1) or SI (Q2) and encompasses a few variables, including accuracy and response time in both short and long terms. However, in the discussion section, we have discussed the findings in relation to accuracy and response time separately but for PI and SI both together. Therefore, while we have discussed all the findings regarding all research questions, referring back to each research question separately would disrupt the conciseness and effectiveness of our discussion due to the way we have discussed the findings. However, we agree that reminding the readers of the research questions at the beginning of the discussion would be helpful. Therefore, we have now mentioned what the research questions asked at the beginning of the discussion.

• Comment: Also can you explain in the discussion more explicitly whether your results and why your results support Van Patten’s model? It would be helpful to talk about implicit/explicit and cognitive processing effects of PI/SI.

Response: Comment accommodated! We have explained this in the first paragraph of the discussion section.

• Comment: Can you discuss other possible reasons why you didn’t find a difference in response times? It may be low power, but what if it isn’t? What does that say about PI/SI instruction?

Response: We have now expanded the discussion to consider additional possible reasons for the lack of difference in response times, including the influence of the participants’ L1 and the relatively shorter instructional period compared to previous studies.

Reviewer 3 Report

Comments and Suggestions for Authors

The article addresses a highly interesting topic. The research method is well designed, and the data are clearly presented. However, there are several issues that should be addressed to improve the article’s clarity and strengthen the authors’ claims. This paper has great potential, and with some revisions, it could become an important contribution to the field.

1. One major concern lies in Section 2.2, where more detailed explanations—and especially examples—of PI are needed. It is not clear how PI and SI differ from each other. In lines 127–129, the authors state that “SI activities are indeed the key component of PI,” which implies that SI is a subset of PI. This is also suggested by line 120, where PI is described as comprising Processing and Instruction, again reinforcing the idea that SI is a component of PI. While this relationship is not problematic per se, it does become confusing when two separate experimental groups (PI and SI) are created, without a clear early distinction. The confusion is compounded by several references throughout the paper to “PI/SI”, which further blurs the distinction and makes it difficult to understand the rationale for treating them as two distinct experimental conditions. Although the authors finally clarify in lines 338–343 that the difference lies in the fact that the PI group received instruction on the processing component while the SI group did not, this distinction is not made sufficiently clear earlier in the paper. Moreover, the initial presentation of PI lacks examples, making this clarification less effective. This issue also affects the interpretation of RQ3. Since the research questions are presented after a literature review that tends to group PI and SI together, it is unclear why we should expect different outcomes for these two groups in the first place.
2. Relatedly, the data analysis shows no significant difference between the PI and SI groups. However, the authors do not attempt to offer any explanation for this. Could it be because PI and SI are not meaningfully distinct in practice? Based on the literature review—which frequently treats PI and SI as overlapping—it would not be surprising to find no significant difference. Still, the authors should at least attempt a brief explanation or propose this as a point for future research. As it stands, the lack of differentiation in the literature review may bias the reader’s interpretation and obscure the implications of this finding.
3. My main concern regards the statistical model employed. Although the authors correctly test for sphericity and implement adjustments where needed, sphericity is not the only assumption underlying ANOVA. The use of parametric tests such as ANOVA could lead to false positives or false negatives. Non-parametric alternatives might be more appropriate and would likely strengthen the validity of the results. This is especially relevant in the case where the only significant difference in reaction time is between SI and Control in the intermediate posttest—a result that may be a false negative if the underlying data do not meet parametric assumptions.

Other minor revisions:

• There are occasional font inconsistencies (e.g., lines 79–81, 147–149, 153, 190–199), which should be standardized.
• In the abstract, the acronym PI appears before being defined. It should be introduced in full at first mention.
• Please double-check editorial guidelines regarding citation format. You write “(Benati, 2022a, b)”, but “(Benati, 2022a, 2022b)” is more common in academic publications.
• Line 151: “The production tests were biased for the DG group.” This statement is unclear. What exactly is meant by “biased”? Please clarify.
• Lines 324–325: The claim that passive sentences in Persian typically follow OSV order is contradicted by the example given (which is also not numbered), where the verb appears in initial position.
• Line 364: “Fillers” would be a better term than “distractors” in this context.
• Table 4: The p-values are not readable. The formatting should be adjusted to ensure clarity.

Author Response

• Comment: One major concern lies in Section 2.2, where more detailed explanations—and especially examples—of PI are needed. It is not clear how PI and SI differ from each other. In lines 127–129, the authors state that “SI activities are indeed the key component of PI,” which implies that SI is a subset of PI. This is also suggested by line 120, where PI is described as comprising Processing and Instruction, again reinforcing the idea that SI is a component of PI. While this relationship is not problematic per se, it does become confusing when two separate experimental groups (PI and SI) are created, without a clear early distinction. The confusion is compounded by several references throughout the paper to “PI/SI”, which further blurs the distinction and makes it difficult to understand the rationale for treating them as two distinct experimental conditions. Although the authors finally clarify in lines 338–343 that the difference lies in the fact that the PI group received instruction on the processing component while the SI group did not, this distinction is not made sufficiently clear earlier in the paper. Moreover, the initial presentation of PI lacks examples, making this clarification less effective. This issue also affects the interpretation of RQ3. Since the research questions are presented after a literature review that tends to group PI and SI together, it is unclear why we should expect different outcomes for these two groups in the first place.

Response: We have now further clarified the distinction between PI and SI at the end of Section 2.2. Also, as we have now explained, the examples provided for SI hold true for PI, too. The only difference between SI and PI is the provision of explicit information in the latter.

There is now ample evidence showing that PI and SI are equal in their effects on improving L2 learners’ accuracy, which suggests that explicit information does not provide any additional benefits in terms of accuracy. This evidence, however, is based on studies using offline measures. No studies have examined whether the provision of explicit information would offer any additional benefits in areas other than accuracy. Therefore, we treated SI and PI as two different experimental groups in this study. The aim was to examine whether there would be a significant difference between PI and SI not only in accuracy but more importantly in response time.

• Comment: Relatedly, the data analysis shows no significant difference between the PI and SI groups. However, the authors do not attempt to offer any explanation for this. Could it be because PI and SI are not meaningfully distinct in practice? Based on the literature review—which frequently treats PI and SI as overlapping—it would not be surprising to find no significant difference. Still, the authors should at least attempt a brief explanation or propose this as a point for future research. As it stands, the lack of differentiation in the literature review may bias the reader’s interpretation and obscure the implications of this finding.

Response: Comment accommodated! We have explained this at the end of the first paragraph of the discussion.

• Comment: My main concern regards the statistical model employed. Although the authors correctly test for sphericity and implement adjustments where needed, sphericity is not the only assumption underlying ANOVA. The use of parametric tests such as ANOVA could lead to false positives or false negatives. Non-parametric alternatives might be more appropriate and would likely strengthen the validity of the results. This is especially relevant in the case where the only significant difference in reaction time is between SI and Control in the immediate posttest—a result that may be a false negative if the underlying data do not meet parametric assumptions.

Response: Thank you for highlighting this important issue. We double-checked the data analysis results and noticed that the assumptions underlying ANOVA have not been violated. Therefore, we do not think running the non-parametric test would be appropriate. However, we have now removed the control group from the response time data analysis. As we have now explained in the results section, the control group did not demonstrate any improvement in accuracy on passive sentences across the posttests. Therefore, response time was not analysed for this group, as it would not yield meaningful information about changes in processing efficiency in the absence of learning.

Other minor revisions:

• Comment: There are occasional font inconsistencies (e.g., lines 79–81, 147–149, 153, 190–199), which should be standardized.

Response: Comment accommodated!

• Comment: In the abstract, the acronym PI appears before being defined. It should be introduced in full at first mention.

Response: Comment accommodated!

• Comment: Please double-check editorial guidelines regarding citation format. You write “(Benati, 2022a, b)”, but “(Benati, 2022a, 2022b)” is more common in academic publications.

Response: Comment accommodated!

• Comment: Line 151: “The production tests were biased for the DG group.” This statement is unclear. What exactly is meant by “biased”? Please clarify.

Response: It means that the production tests gave an unfair advantage to the DG group over the PI group as the DG group received output practice during the instruction, while the PI group did not. We have now clarified this in the manuscript.

• Comment: Lines 324–325: The claim that passive sentences in Persian typically follow OSV order is contradicted by the example given (which is also not numbered), where the verb appears in initial position.

Response: Unlike English, Persian is written from right to left. Therefore, the word-to-word English translation provided for the Persian sentence in the example is to be read from right to left. To avoid confusion, we have now added that Persian is written from right to left.

• Comment: Line 364: “Fillers” would be a better term than “distractors” in this context.

Response: Comment accommodated!

• Comment: Table 4: The p-values are not readable. The formatting should be adjusted to ensure clarity.

Response: Table 4 revised!

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Most of my comments have been addressed in this revision but some comments need further clarification, and one comment raised concerns about scientific adequacy and publishing ethics.

1. The revisions are hard to process – it’s not clear what the text highlighted in red replaces, which makes it hard to know whether the new text is improved.
2. I am concerned that you have data that would more adequately address your research question but choose not to use it in this paper. I would strongly recommend strengthening the current paper with the word-by-word reaction time data and then revising your plan for the next paper. Otherwise, a very strong rationale should be supplied for why this paper contributes anything useful. 
3. In response to this original comment “p. 6, l. 278: Some speculation about what might have caused the different results in Lee et al 2020 vs Malorvrh et al. 2020 would be interesting here – e.g. were L1s different and in which ways?”, the author(s) state that they cannot find any differences to account for the mixed findings from the two studies. However, this information should be added to the paper – it should be made clear, too, to be convincing for the reader, on which parameters the two studies were the same.
4. I don’t understand this response (what does it mean that the ‘control would only take tests’?), and please explain where in the text any changes have been made to address the original comment: Exposure to the passive structure (although without instruction) would indeed count as a type of instruction (input flood). Furthermore, previous research has indicated that mere exposure to the target structure does not result in any significant achievements in sentence interpretation (e.g., ChiuChiù & Benati, 2020; Issa & Morgan-Short, 2019). Therefore, following similar studies (e.g., Benati, 2023; Benati & Batziou, 2017; Benati & Chan, 2023; Cadierno, 1995; Cheng, 2004; Lee, 2015; VanPatten & Wong, 2004), a decision was made that the control would only take tests.

Comments on the Quality of English Language

The way revisions were made (which may be a submission system flaw, I can't tell) makes it very hard to judge the coherence of the text, so I would encourage the authors to double-check everything to ensure that the text flows and is coherent and easily readable. 

Author Response

Comment 1: The revisions are hard to process – it’s not clear what the text highlighted in red replaces, which makes it hard to know whether the new text is improved.

Response: We used the “Track Changes” mode to highlight the revisions (additions and deletions). We are not sure what the text looks like after submitting the manuscript. This might be a technical issue. We have now provided a clean version for ease of reading.

Comment 2: I am concerned that you have data that would more adequately address your research question but choose not to use it in this paper. I would strongly recommend strengthening the current paper with the word-by-word reaction time data and then revising your plan for the next paper. Otherwise, a very strong rationale should be supplied for why this paper contributes anything useful.

Response: We appreciate the comment and agree that the word-by-word reaction time data can provide further information, as acknowledged in the manuscript. However, as we explained before, we plan to include this data in another paper. We believe including the word-by-word data in our next paper would work better as, in that paper, we will focus on word-by-word reaction time based on input modality (i.e., written vs. auditory). Additionally, the research questions we have proposed in this study focus on accuracy of response and response time, not reading time. As such, the data we have provided answer the research questions, and including reading time data would be beyond the scope of this study. Finally, as explained at the end of Section 2.5, we believe the significance of the study and its contribution to the field mainly lies in its investigation of the long-term effects of PI and SI rather than the inclusion of reading time data. That said, we still agree that word-by-word reaction time can provide additional information. However, providing this data is beyond the scope/aim of the study, and we believe the study contributes to the field through investigating the long-term effects of instruction as measured by not only accuracy but also response time in picture selection.

Comment 3: In response to this original comment “p. 6, l. 278: Some speculation about what might have caused the different results in Lee et al 2020 vs Malorvrh et al. 2020 would be interesting here – e.g. were L1s different and in which ways?”, the author(s) state that they cannot find any differences to account for the mixed findings from the two studies. However, this information should be added to the paper – it should be made clear, too, to be convincing for the reader, on which parameters the two studies were the same.

Response: We have indeed provided this information in the manuscript in Section 2.5 (para. 7) and Section 5 (Para. 2).

Comment 4: I don’t understand this response (what does it mean that the ‘control would only take tests’?), and please explain where in the text any changes have been made to address the original comment: Exposure to the passive structure (although without instruction) would indeed count as a type of instruction (input flood). Furthermore, previous research has indicated that mere exposure to the target structure does not result in any significant achievements in sentence interpretation (e.g., ChiuChiù & Benati, 2020; Issa & Morgan-Short, 2019). Therefore, following similar studies (e.g., Benati, 2023; Benati & Batziou, 2017; Benati & Chan, 2023; Cadierno, 1995; Cheng, 2004; Lee, 2015; VanPatten & Wong, 2004), a decision was made that the control would only take tests.

Response: We apologize for the confusion caused by our response. The original comment asked how we can make sure that the difference between the control and treatment groups is due to the instructional treatment and not merely additional exposure to the target structure. The concern, then, was that the control group should have been offered a corresponding amount of exposure to passives (but without instruction). Our response was that providing any instruction to the control group, even in the form of only exposure to the target structure, would actually count as a type of instructional treatment, which, therefore, would cause the group not to be control any more. Therefore, the control group did not receive any instruction and only took the tests (pretest, immediate posttest, and delayed posttest). This is actually a common practice in PI literature (e.g., Benati, 2023; Benati & Batziou, 2019; Benati & Chan, 2023; Cadierno, 1995; Cheng, 2004; VanPatten & Wong, 2004).

Furthermore, previous research has indicated that only exposure to the target structure does not result in any significant achievements in sentence interpretation (e.g., ChiuChiù & Benati, 2020; Issa & Morgan-Short, 2019). Therefore, the observed effects for the treatment groups are not likely to be due to merely exposure to the target structure. However, we agree that providing some explanations in this regard would be helpful. Therefore, we have now added a note explaining this point.

Comment 5: The way revisions were made (which may be a submission system flaw, I can't tell) makes it very hard to judge the coherence of the text, so I would encourage the authors to double-check everything to ensure that the text flows and is coherent and easily readable.

Response: We have double-checked the manuscript in terms of language. Additionally, we have now provided a clean version of the manuscript as well.

Reviewer 3 Report

Comments and Suggestions for Authors

No further comments.

Author Response

Many thanks for your time in reviewing the manuscript again.