The Role of Determiners in the Processing of Gender Agreement Morphology by Heritage Speakers of Spanish
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsReview of: The role of determiners in the processing of gender agreement morphology by heritage speakers of Spanish
Abstract
page1. Since this study is largely a replication of Keating’s (2022, 2024) work, delete “conceptually” in “conceptually replicated.” Also, instead of “comparison group,” use “control group for comparison” (here and elsewhere in the manuscript).
- Introduction
page 1. To be consistent with the other examples, add the word “match” or change to “gender match” in example (1)a.
page 2. First paragraph. Add a brief definition, within parenthesis (see page 5), when first introducing a concept. In this case, for “non-local agreement.” This clarifies from the start how the author is using a term, and thereby helps to avoid misunderstandings. For example, in previous studies (e.g., Keating, 2009), “non-local agreement” is defined as outside of the noun phrase, but within the verb phrase. The current author(s) refers to “non-local agreement” as non-adjacent to the head noun. Using adjacent vs. non-adjacent is more accurate for the current study.
page 2. Add a citation here. The author suggests that the current study is going to test how a gender-marked determiner might act as a gender cue. This has already been tested in previous studies, for example, Lew-Williams et al. (2010).
Reference: Lew-Williams, C., & Fernald, A. (2010). Real-time processing of gender-marked articles by native and non-native Spanish speakers. Journal of Memory and Language, 63, 447-464.
page 2. The introduction should provide information on what has been done and what we know so far about the issue, and then explain why the current study is interesting and what the contribution to the current literature will be. This is missing here.
page 2-7. The next subsections should be an independent Review of the literature, rather than a part of the Introduction.
1.1 Gender in Spanish
page 3. First paragraph. Add a citation to the last sentence supporting the claim that age is a factor in gender agreement variation for heritage speakers. Also, add a sentence or two to expand/justify the focus on heritage speakers.
1.2 Age effects HSs
page 3, paragraph 3. Briefly define “macroparametric and microparametric features” when first introduced.
page 4, first paragraph. In order to effectively contrast Cuza & Tattam’s study with Montrul & Potowski’s, include the children’s ages in both.
page 4, last paragraph. By “highly proficient HSs” do you mean HSs with 7-9 AOB? Clarify.
1.3 Linear Distance Effects
page 5, first paragraph. Within parentheses, briefly define “structural distance,” “linear distance,” and “ERP artifacts.”
page 5, paragraph 3. Add the Lew-Williams & Fernald’s (2010) citation here (provided above). The question the author asked at the beginning of the paragraph has been addressed.
1.4 Cue-Based Processing
page 5, last paragraph. Briefly define “P600.”
page 7, top. Add a citation here. There have been studies, with both Spanish L2 and HSs, that have found that a gender-marked determiner affects gender agreement processing within the same noun phrase. For example, Hopp & Lemmerth (2018), as well as other authors cited elsewhere in the current study.
Reference: Hopp, H., & Lemmerth, N. (2018). Lexical and syntactic congruency in L2 predictive gender processing. Studies in Second Language Acquisition, 40, 171-199.
1.5 This study
page 7. Add “in HSs” to the question “Does AOB play a role?” in both RQ1 and RQ2.
page 7, second paragraph. Here (and elsewhere) delete “conceptually” in “conceptually replicates.” Again, this is primarily a replication study.
page 7, third paragraph. Define how “proficiency” is used in the current study. Also, add a citation after “ … determiners – acquired early in L1 acquisition- serve as strong cue… .” In addition, after stating the expected effects of AOB in the predictions add a citation: this finding has been found in previous studies.
- Materials and Methods
2.1 Participants
page 7, bottom. Delete information about the original pool of participants, which is not relevant. But add the number of male participants next to the number of female participants when describing both the HSs and the control groupage Also, provide the number of undergraduate students participating as well as the number of graduate students. In addition, explain how the participants were recruited.
page 8. Top paragraph. Are the late L2 English acquirers post-pubescent learners of English? Clarify.
page 8, TABLE 1. The HSs are divided into three groups, based on their age of first exposure to English. The self-reports and DELE test results do not show large variation between these three groups, however, so without some descriptive statistical testing (e.g., t-tests) it is not possible to know whether these three groups are significantly different from each other. This is essential information.
page 9. TABLE 1. The dominance score (BLP) shows that the 4-6 HSs group is more Spanish dominant than the 7-9 groupage The author needs to provide an explanation for this counter-intuitive, and potentially important, observation.
2.2 Materials
page 9. Specify the study indicated by “… used in the original study”? Also, if the intention were to use a “longer” modifier between the noun and the adjective, why not use “algunos/as” instead of “unos/unas”? The justification for using an indefinite rather than definite determiner in the experimental sentences is unclear and irrelevant: “[they] allow DPs to serve as direct objects of transitive verbs….” Either change or delete.
page 10, first paragraph. The experimental nouns, as well as the adjectives, vary in length from two to four syllables. What’s the rationale for not having the same length in all target nouns and adjectives? In addition, provide examples of distractors and fillers used in the current study. Also, each participant was exposed to 32 experimental items, 32 distractors, and 64 fillers, but were participants also asked to respond to128 comprehension questions? Clarify.
2.3 General Procedure
page 10, first paragraph. Were the participants tested individually? In groups? Make this explicit. Also, what is meant by “when drifting occurred?” Give an example, and how often it happened.
page 10, second paragraph. Add DELE to “the proficiency test.” How long did the reading task take participants to complete? Also, include the rationale for starting the study procedure with the experimental reading test BEFORE the memory experiment (not included in this current manuscript), the language background questionnaire, the DELE test, the BLP test, and the vocabulary task. Why this sequence?
2.4 Data Analysis
page 11, second paragraph. Since the adjectives have three different lengths: two-, three-, and four-syllables, the effects of adjective length should have been included in the current study. Why weren’t they? Also, is the “target word test” part of the vocabulary test, or is it a different test that was not described earlier? Clarify. Also, be more precise regarding what will be analyzed. For example, in this paragraph, we read that “the effects of gender agreement violation” will be examined, but in the next section, Results, the wording is different: “the effects of Grammaticality” (page 14). Are these the same? It would be helpful to include a list of exactly what is to be measured, to label these variables consistently, and to explain precisely what is meant by “grammaticality.”
- Results
pages 15-17. The interaction between proficiency and the AOB 0-3 and AOB 4-6 groups was reported, but not the interaction with the AOB 7-9 groupage Also, there is no reported comparison between groups AOB 4-6 and AOB 7-9. Why were these excluded?
page 18. Was a different analysis performed in two “new” groups: AOB 0-6 and AOB 7-9? If so, please explain. (See top paragraph and figure 14.) There is no mention of AOB 7-9 when commenting on the No Cue condition results taken together (last paragraph).
page 19. Revise the age numbers of the groups (top paragraph). Are there now three groups with different AOBs; 0-6, 4-9, 7-10; than at the beginning of the manuscript? Please EXPLAIN.
- Discussion
4.1 Delayed sensitivity to gender agreement violations at a distance
page 19. This section requires significant elaboration and expansion. For example, the result that the group with the highest AOB, 7-9, did not display the highest proficiency, was not discussed at all, nor was the role of Spanish dominance. Since these findings are not completely consistent with Keating (2022, 2024), they should be discussed in this section. Also, issues including language experience, memory, proficiency, and metalinguistic awareness, are raised, but given insufficient attention.
4.2 Determiners as cues for agreement processing
page 20. This section also needs more discussion. And more citations, since the current results support numerous previous findings. For example, many existing research findings have provided evidence that gender-marked determiners facilitate morphosyntactic processing for L1, L2, and HSs.
4.3 The role of AOB: Early and late effects
page 20. This section also needs to be substantially expanded. Why is there no mention of the AOB 7-9 group? Also, without statistical testing for significant differences in proficiency between the three AOB groups, the claim that proficiency increases with a delay in the onset of bilingualism is not sustained. A thorough revision is necessary here.
4.4 Theoretical implications and future directions
page 20. Add a citation for the claim that “heritage speakers are not necessarily impaired in morphosyntactic processing….” Also, some of the issues proposed, such as “experience with heritage language schooling,” might explain why the AOB 4-6 group was more Spanish dominant than the AOB 7-9 groupage This issue needs to be addressed.
- Conclusion
page 21. Although this is an adequate summary of the results, since there is a need for major revisions in the manuscript, the final Conclusion section will need to be revised as well.
Author Response
Thank you for taking the time to review our manuscript. Find our responses below:
Comment: page1. Since this study is largely a replication of Keating’s (2022, 2024) work, delete “conceptually” in “conceptually replicated.” Also, instead of “comparison group,” use “control group for comparison” (here and elsewhere in the manuscript).
Response: We believe our study is a conceptual replication of Keating (2022; 2025) by testing the same core hypothesis – namely, the processing of gender agreement morphology at a distance– under new methodological conditions. In line with McManus’ (2024) definition, conceptual replication involves three or more major changes from the original study. We introduced (1) a new set of stimulus sentences using the same nouns and adjectives, (2) an additional condition incorporating a gender cue to examine its impact on the time course of agreement processing, and (3) a new participant population of heritage speakers in the Midwest as opposed to Southern California. These changes test the generalizability of the original findings across stimuli, conditions, and populations, while maintaining alignment with the original theoretical framework.
McManus, K. (2024). Replication studies in second language acquisition research: Definitions, issues, resources, and future directions: Introduction to the special issue. Studies in Second Language Acquisition, 46(5), 1299–1319. https://doi.org/10.1017/S0272263124000652
Finally, we maintain that our monolingually-raised group is not a control group but a comparison group. This is motivated by our contrast coding approach, taken from Keating (2022; 2025), that splits participants by AOB subgroups. Naturally, the comparison group was exposed to English post-puberty, which makes them the ideal comparison group.
Introduction
Comment: page 1. To be consistent with the other examples, add the word “match” or change to “gender match” in example (1)a.
Response: Thank you for pointing this out. We agree. Therefore, we have made this change.
Comment: page 2. First paragraph. Add a brief definition, within parenthesis (see page 5), when first introducing a concept. In this case, for “non-local agreement.” This clarifies from the start how the author is using a term, and thereby helps to avoid misunderstandings. For example, in previous studies (e.g., Keating, 2009), “non-local agreement” is defined as outside of the noun phrase, but within the verb phrase. The current author(s) refers to “non-local agreement” as non-adjacent to the head noun. Using adjacent vs. non-adjacent is more accurate for the current study.
Response: Thank you for pointing this out. We have added a clarifying definition in parentheses at first mention: “non-local agreement” (e.g., when a word intervenes between the head noun and its adjective). While we recognize that Keating (2009) defines “non-local” in relation to NP vs. VP boundaries, our study uses the term to refer specifically to linear distance between agreeing elements—namely, head nouns and adjectives. We chose not to adopt the adjacent/non-adjacent terminology used in Keating (2022; 2024) to avoid implying the presence of an adjacent (i.e., local) baseline condition in our design, which we do not include. The revised phrasing is intended to prevent confusion while aligning more accurately with the structure of our stimuli.
Comment: page 2. Add a citation here. The author suggests that the current study is going to test how a gender-marked determiner might act as a gender cue. This has already been tested in previous studies, for example, Lew-Williams et al. (2010).
Response: Thank you for this important comment. We have now cited relevant prior work (e.g., Lew-Williams & Fernald, 2010; Grüter et al., 2012; Hopp, 2013; Fuchs, 2022a, 2022b) that demonstrates how gender-marked determiners can serve as morphosyntactic cues, particularly in auditory processing contexts using the visual world paradigm. However, most of these studies focus on L1/L2 speakers, and only Fuchs (2022a, 2022b) investigates heritage speakers. More crucially, these studies examine auditory predictive processing, where determiners function as anticipatory cues for upcoming nouns. Our study, by contrast, focuses on agreement processing during reading, specifically when agreement is non-local and cognitively demanding. We argue that in such contexts, earlier cues like determiners may facilitate gender feature activation or retrieval, which in turn may enhance the detection of gender agreement violations. To our knowledge, such mechanism has not been empirically tested in the reading modality. To clarify this distinction while acknowledging prior work, we have added citations and contextualized our contribution at the end of the introduction.
Reference: Lew-Williams, C., & Fernald, A. (2010). Real-time processing of gender-marked articles by native and non-native Spanish speakers. Journal of Memory and Language, 63, 447-464.
Comment: page 2. The introduction should provide information on what has been done and what we know so far about the issue and then explain why the current study is interesting and what the contribution to the current literature will be. This is missing here.
Response: Thank you for pointing this out. We agree. Therefore, we have rewritten the last paragraph in the introduction to address these concerns.
Comment: page 2-7. The next subsections should be an independent Review of the literature, rather than a part of the Introduction.
Response: Thank you for pointing this out. We have now changed this section to be a literature review.
1.1 Gender in Spanish
Comment: page 3. First paragraph. Add a citation to the last sentence supporting the claim that age is a factor in gender agreement variation for heritage speakers. Also, add a sentence or two to expand/justify the focus on heritage speakers.
Response: Thank you for pointing this out. We have now added the corresponding citations, and a sentence justifying the focus on HSs.
1.2 Age effects HSs
Comment: page 3, paragraph 3. Briefly define “macroparametric and microparametric features” when first introduced.
Response: Thank you for pointing this out. We have now added definitions in parentheses.
Comment: page 4, first paragraph. In order to effectively contrast Cuza & Tattam’s study with Montrul & Potowski’s, include the children’s ages in both.
Response: Thank you for pointing this out. We have now added the ages for Cuza & Tattam’s study.
page 4, last paragraph. By “highly proficient HSs” do you mean HSs with 7-9 AOB? Clarify.
Response: Thank you for pointing this out. We have now added AOB ranges in parentheses.
1.3 Linear Distance Effects
Comment: page 5, first paragraph. Within parentheses, briefly define “structural distance,” “linear distance,” and “ERP artifacts.”
Response: Thank you for pointing this out. We have now added definitions in parentheses.
Comment: page 5, paragraph 3. Add the Lew-Williams & Fernald’s (2010) citation here (provided above). The question the author asked at the beginning of the paragraph has been addressed.
Response: Thank you for this helpful suggestion. We agree that previous work—particularly Lew-Williams & Fernald (2010)—has addressed similar questions in the auditory domain. While we maintain that auditory predictive processing may engage different mechanisms than those involved in reading, these studies provide a valuable foundation for our hypothesis. We have revised the question to include the citation and now frame it as: “as evidence from auditory predictive processing suggests (Lew-Williams & Fernald, 2010; Grüter et al., 2012; Hopp, 2013)?” This acknowledges prior findings while clarifying the distinct contribution of our study in the reading modality.
1.4 Cue-Based Processing
Comment: page 5, last paragraph. Briefly define “P600.”
Response: Thank you for pointing this out. We have now added definitions in parentheses
Comment: page 7, top. Add a citation here. There have been studies, with both Spanish L2 and HSs, that have found that a gender-marked determiner affects gender agreement processing within the same noun phrase. For example, Hopp & Lemmerth (2018), as well as other authors cited elsewhere in the current study.
Response: Thank you for this comment. We acknowledge that prior studies—including Hopp & Lemmerth (2018) and Fuchs (2022a)—have shown that gender-marked determiners can influence gender agreement processing within the noun phrase for both L2 and heritage speakers. While much of this evidence comes from auditory studies using the Visual World Paradigm and focuses on predictive processing, we argue that reading involves distinct mechanisms, particularly in cases of non-local agreement. To reflect this, we have added the citation to our discussion of the gap, now phrased as: “as predictive processing studies suggest (e.g., Hopp & Lemmerth, 2018; Fuchs, 2022a).” This better situates our contribution within the broader literature while underscoring the novelty of our reading-based approach.
Reference: Hopp, H., & Lemmerth, N. (2018). Lexical and syntactic congruency in L2 predictive gender processing. Studies in Second Language Acquisition, 40, 171-199.
1.5 This study
Comment: page 7. Add “in HSs” to the question “Does AOB play a role?” in both RQ1 and RQ2.
Response: Thank you for pointing this out. We have now added “in HSs” to both questions.
Comment: page 7, second paragraph. Here (and elsewhere) delete “conceptually” in “conceptually replicates.” Again, this is primarily a replication study.
Response: Thank you for your comment. Although we believe our study is a conceptual replication, we have now clarified this in our intro and used the term “replication” here forth.
Comment: page 7, third paragraph. Define how “proficiency” is used in the current study. Also, add a citation after “ … determiners – acquired early in L1 acquisition- serve as strong cue… .” In addition, after stating the expected effects of AOB in the predictions add a citation: this finding has been found in previous studies.
Response: Thank you for pointing this out. We have now added what we meant by proficiency. We also added citations for the L1 acquisition claims. We also added “found in predictive processing studies in HSs (Fuchs, 2022a).” to the claim related to determiners.
- Materials and Methods
2.1 Participants
Comment: page 7, bottom. Delete information about the original pool of participants, which is not relevant. But add the number of male participants next to the number of female participants when describing both the HSs and the control groupage Also, provide the number of undergraduate students participating as well as the number of graduate students. In addition, explain how the participants were recruited.
Response: Thank you for pointing this out. We have now deleted the information about the initial pool of participants. We also added the number of male participants, and undergraduate and graduate students. We also added a line about recruitment.
Comment: page 8. Top paragraph. Are the late L2 English acquirers post-pubescent learners of English? Clarify.
Response: Thank you for pointing this out. We have now added AOB in English for this group in parentheses.
Comment: page 8, TABLE 1. The HSs are divided into three groups, based on their age of first exposure to English. The self-reports and DELE test results do not show large variation between these three groups, however, so without some descriptive statistical testing (e.g., t-tests) it is not possible to know whether these three groups are significantly different from each other. This is essential information.
Response: Thank you for pointing this out. We have now added ANOVAs for proficiency and dominance. None of analysis were significant between HS groups.
Comment: page 9. TABLE 1. The dominance score (BLP) shows that the 4-6 HSs group is more Spanish dominant than the 7-9 groupage The author needs to provide an explanation for this counter-intuitive, and potentially important, observation.
Response: Thank you for pointing this out. However, the 4-6 HSs group was not significantly more Spanish dominant than the 7-9 groupage. In addition to the ANOVAS, an independent-samples t-test confirmed no significant difference in dominance scores between participants in the 4–6 AOB group (M = 25.61) and those in the 7-9 AOB group (M = 26.08). t (28.77) = 0.04, p = .966, 95% CI [−22.02, 22.96].
2.2 Materials
Comment: page 9. Specify the study indicated by “… used in the original study”? Also, if the intention were to use a “longer” modifier between the noun and the adjective, why not use “algunos/as” instead of “unos/unas”? The justification for using an indefinite rather than definite determiner in the experimental sentences is unclear and irrelevant: “[they] allow DPs to serve as direct objects of transitive verbs….” Either change or delete.
Response: Thank you for pointing this out. We have now specified what study we refer to.
Regarding the suggestion to use algunos/as as a longer modifier, we appreciate this point. However, algunos/as cannot grammatically appear postnominally as modifiers (e.g., árboles *algunos altos), and while they can function as pre-nominal gender cues (e.g., algunos árboles altos), that configuration was not the focus of the current study. We agree that exploring algunos/as as cues could be a promising direction for future work, especially for processing agreement at a longer distance in Keating’s (2009) fashion.
As for our use of unos/as rather than los/las, this decision emerged from the norming phase of the stimuli in which native speaker consultants evaluated sentence naturalness and grammaticality. Our goal was to construct stimuli that were both natural and structurally consistent across cue and no-cue conditions. We found that using unos/as in combination with the intensifier muy yielded the most acceptable sentences in both conditions without requiring changes to the adjective or compromising sentence flow. For instance (1):
(1)
el granjero cortó árboles muy altos en la cima de la montaña à preferred.
el granjero cortó árboles más altos en la cima de la montaña
el granjero cortó los árboles muy altos en la cima de la montaña
el granjero cortó los árboles más altos en la cima de la montañaà preferred.
As evident, the sentence that works best in the no-cue condition contains “muy”, while the sentence in the cue condition that works best contains “más”, resulting in different modifiers across versions of the same sentence. This is not ideal for sentence processing studies. Fortunately, when using “unos” and “muy” across the board, the issue disappears (2):
(2)
el granjero cortó árboles muy altos en la cima de la montaña à preferred.
el granjero cortó unos árboles muy altos en la cima de la montañaà preferred.
Because of this, we decided that the best way to keep items consistent across conditions was to use indefinite determiners and the modifier “muy”. We have now reworded the way we constructed sentences to justify the use of indefinite determiners.
Comment: page 10, first paragraph. The experimental nouns, as well as the adjectives, vary in length from two to four syllables. What’s the rationale for not having the same length in all target nouns and adjectives? In addition, provide examples of distractors and fillers used in the current study. Also, each participant was exposed to 32 experimental items, 32 distractors, and 64 fillers, but were participants also asked to respond to128 comprehension questions? Clarify.
Response: Thank you for pointing this out. The experimental nouns and adjectives were taken from Keating (2022; 2025). We decided to keep these for comparability across studies.
We have now added an example of the distractors and fillers and clarified that all participants were asked comprehension questions after each sentence, whether it was experimental, distractor, or filler.
2.3 General Procedure
Comment: page 10, first paragraph. Were the participants tested individually? In groups? Make this explicit. Also, what is meant by “when drifting occurred?” Give an example, and how often it happened.
Response: Thank you for pointing this out. We have now added “individually” to make it explicit that participants were tested as such, one by one. We also added an explanation of what we meant by “drifting.”
As per how often this happened, we are unfortunately unable to answer that. Drifts, or the deviation of the eye gaze recording upwards or downwards from the sentence, may occur due to multiple reasons: a participant is wearing reading glasses that drop overtime, the participant is wearing contact lenses, or the participant is wearing makeup that tricks the eye tracker to think the eyes are looking somewhere else. It may also happen because participants moved their heads, or simply because the participant is tired. Drifting may occur consistently for a participant, or not at all for others. This is why eye tracking researchers must be alert throughout the experiment to identify drifting, and perform calibration as needed. The data viewer software allows researchers to drift correct sentences, following the recommendation in Godfroid (2020, p. 257). we do not have data on how often drift occurred in our study because it is not easily quantified, reason why this is not typically reported (Godfroid et al., 2025, p- 278).
Godfroid, A. (2019). Eye Tracking in Second Language Acquisition and Bilingualism: A Research Synthesis and Methodological Guide (1st ed.). Routledge. https://doi.org/10.4324/9781315775616
Godfroid, A., Finch, B., and Koh, J. (2025), Reporting Eye-Tracking Research in Second Language Acquisition and Bilingualism: A Synthesis and Field-Specific Guidelines. Language Learning, 75(1). 250-294. https://doi.org/10.1111/lang.12664
Comment: page 10, second paragraph. Add DELE to “the proficiency test.” How long did the reading task take participants to complete? Also, include the rationale for starting the study procedure with the experimental reading test BEFORE the memory experiment (not included in this current manuscript), the language background questionnaire, the DELE test, the BLP test, and the vocabulary task. Why this sequence?
Response: Thank you for pointing this out. We have now added DELE to the proficiency test, and the time participants took in the reading task. We also added a line explaining why we performed the reading task first, and the reason behind our sequence.
2.4 Data Analysis
Comment: page 11, second paragraph. Since the adjectives have three different lengths: two-, three-, and four-syllables, the effects of adjective length should have been included in the current study. Why weren’t they? Also, is the “target word test” part of the vocabulary test, or is it a different test that was not described earlier? Clarify. Also, be more precise regarding what will be analyzed. For example, in this paragraph, we read that “the effects of gender agreement violation” will be examined, but in the next section, Results, the wording is different: “the effects of Grammaticality” (page 14). Are these the same? It would be helpful to include a list of exactly what is to be measured, to label these variables consistently, and to explain precisely what is meant by “grammaticality.”
Response Adjectives wew identical across conditions and grammaticality versions of the sentence, meaning that the same 16 adjectives used in the grammatical/ungrammatical versions of the stimuli in the cue condition were the same 16 adjectives. As such, we included adjective length as a covariate, consistent with Keating (2022; 2025).Because of this, we believe the effects of adjective length are out of the scope of this paper, as no relevant interaction nor any of its effects contribute to the research questions at hand. We have added an explanation about this to the manuscript.
We have also changed the target word test to “vocabulary test” for consistency. We also clarified that the effect of grammaticality refers to gender agreement match/mismatch at the adjective word.
- Results
Comment: pages 15-17. The interaction between proficiency and the AOB 0-3 and AOB 4-6 groups was reported, but not the interaction with the AOB 7-9 groupage Also, there is no reported comparison between groups AOB 4-6 and AOB 7-9. Why were these excluded?
Response: We have added an explanation of what contrast a specific group effect refers to in parenthesis
Comment: page 18. Was a different analysis performed in two “new” groups: AOB 0-6 and AOB 7-9? If so, please explain. (See top paragraph and figure 14.) There is no mention of AOB 7-9 when commenting on the No Cue condition results taken together (last paragraph).
Response: We have added an explanation of what contrast a specific group effect refers to in parenthesis, and a reminder to readers about our contrast coding approach at the beginning of each result subsection.
Comment: page 19. Revise the age numbers of the groups (top paragraph). Are there now three groups with different AOBs; 0-6, 4-9, 7-10; than at the beginning of the manuscript? Please EXPLAIN.
Response: Thank you for pointing this out. We have now changed it to 7-9.
- Discussion
4.1 Delayed sensitivity to gender agreement violations at a distance
Comment: page 19. This section requires significant elaboration and expansion. For example, the result that the group with the highest AOB, 7-9, did not display the highest proficiency, was not discussed at all, nor was the role of Spanish dominance. Since these findings are not completely consistent with Keating (2022, 2024), they should be discussed in this section. Also, issues including language experience, memory, proficiency, and metalinguistic awareness, are raised, but given insufficient attention.
Response: Thank you for pointing this out. We have rewritten this section to focus solely on the delayed sensitivity from the lens of cue-based retrieval mechanisms. We have left the discussion of individual differences among HSs to section 5.3 (formerly known as 4.3). In that section, we compare our results with Keating’s, and argue that our results may stem from other individual factors, such as schooling instead of proficiency and dominance given that the latter were not significantly difference from each other.
4.2 Determiners as cues for agreement processing
Comment: page 20. This section also needs more discussion. And more citations, since the current results support numerous previous findings. For example, many existing research findings have provided evidence that gender-marked determiners facilitate morphosyntactic processing for L1, L2, and HSs.
Response: We have rewritten this section to include previous research on determiners aiding in morphosyntactic processing, and our interpretation of our results.
4.3 The role of AOB: Early and late effects
Comment: page 20. This section also needs to be substantially expanded. Why is there no mention of the AOB 7-9 group? Also, without statistical testing for significant differences in proficiency between the three AOB groups, the claim that proficiency increases with a delay in the onset of bilingualism is not sustained. A thorough revision is necessary here.
Response: We have rewritten this section as well. We split the results based on cue conditions, and discuss the role of AOB in each condition. We also provide an interpretation for each finding, adding more individualized information regarding schooling experience (Table 3) taken from our language background questionnaire, including the AOB 7-9 group.
4.4 Theoretical implications and future directions
Comment: page 20. Add a citation for the claim that “heritage speakers are not necessarily impaired in morphosyntactic processing….” Also, some of the issues proposed, such as “experience with heritage language schooling,” might explain why the AOB 4-6 group was more Spanish dominant than the AOB 7-9 groupage This issue needs to be addressed.
Response: We have deleted the claim that “heritage speakers are not necessarily impaired in morphosyntactic processing….” Given that the focus of our study is to understand the role of determiners in gender agreement processing by heritage speakers, not necessarily whether they can process or not. When it comes to schooling, we discussed this in the previous section.
- Conclusion
Comment: page 21. Although this is an adequate summary of the results, since there is a need for major revisions in the manuscript, the final Conclusion section will need to be revised as well.
Response: We have now changed the conclusion section accordingly.
Reviewer 2 Report
Comments and Suggestions for AuthorsSee comments in the document.
Comments for author File: 
Comments.pdf
Author Response
Thank you for taking the time to review our manuscript. Find our responses below:
Literature review:
Comment: Although the structures may be different, the authors may benefit from the study by González Alonso et al 2021 as gender cues as well as gender agreement are discussed and they use, among various experiments, eyetracking data were collected (see reference below).
Response: Thank you for providing such a relevant reference. It is indeed a great contribution to cue-based models of sentence comprehension applied to HSs. We added this reference to our discussion section (5.2)
Methodology section:
Comment: I recommend that the authors include a table in the materials section outlining the experimental conditions and example sentences. This will help readers better understand the design and later the analyses.
Response: Thank you for your suggestion. We have now added a table with a sample sentence with each condition laid out.
Comment: Also, when describing the design of the stimuli, I suggest that the authors include information about the animacy of the nouns. Additionally, it is recommended to indicate whether they controlled for the length of the sentence as well as for the experimental words (nouns and adjectives). I know that adjective length was taken into consideration as it is included in the analyses as one of the fixed effects and that it is not discussed in the results section as it is out of the scope of this study. However, I believe that for the sake of clarity, it is important that they discuss it in the description of the design.
Response: We have now added information about the animacy of the nouns, and the range of sentence length in our study, the length of the experimental words, and log10 frequency.
Results:
Comment: On page 18, the authors say: To explore the three-way interaction within each group, two LMM were fitted with Grammaticality and Proficiency as fixed factors predicting sensitivity in total time for each AOB group (0-6 and 7-9). However, in previous section, they always indicate three different groups: 0-3 (simultaneous) and 4-6 (early sequential). Why are early sequential and simultaneous bilinguals in the same group? Could you please justify why in this case simultaneous and early sequential bilinguals are included within the same group?
Response: We addressed this concern to another reviewer. We copy our response.
Recall that we used reversed Helmert contrast for Group coding for our analysis (section 3.4, second paragraph). Group 3 compares simultaneous (AOB 0-3) to early sequential (AOB 4-6) only. Group 2 compares late AOB (7-9) to all earlier AOB (0-6), in the same way Group 1 compares all late AOB (monolingually raised) to all HSs (AOB 0-9). We followed Keating’s method and logic for this analysis that was very useful at understanding the nuances of agreement processing among HSs based on age in a reversed manner, which allows for a comparison of late AOB 7-9 to all earlier ones (0-6). We have added an explanation of what contrast a specific group effect refers to in parenthesis, and a reminder to readers about our contrast coding approach at the beginning of each result subsection
Comment: Minor errors:
- There are some inconsistencies in the format of the in-text references. Sometimes the last names of three authors appear (e.g., Kupisch, Akpinar & Stöhr, 2013, on page 3) while other times the et al. replaces them (Domínguez et al., 1999).
- Spelling error: Monrul, 2008 should be Montrul, 2008.
- Keating’s proposal sometimes appears as 2024 and other times as 2025 – in the reference list is 2025.
Response: Thank you for pointing these out to us. We have changed this.
References:
González Alonso, Jorge, Ian Cunnings, Hiroki Fujita, David Miller and Jason Rothman. 2021. Gender attraction in sentence comprehension. Glossa: a journal of general linguistics 6(1): 20. 1–26. DOI: https://doi.org/10.5334/ gjgl.1300
