Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study employed eye-tracking methodology to investigate the relationship between contextual predictability and readers' segmentation preferences during Chinese sentence reading. The key findings indicated that: (1) contextual predictability did not significantly influence any eye-movement measures; (2) segmentation type produced consistent effects across all measures, with shorter reading times observed for AB-C compared to A-BC segmentation, suggesting a stable preference for two-character segmentation; and (3) no significant interaction emerged between contextual predictability and segmentation type. The authors concluded that these findings demonstrate a robust preference for AB-C segmentation in Chinese reading and suggest that contextual predictability and word segmentation operate as independent processes, with predictability exerting minimal influence on segmentation during reading.

The manuscript is generally well-written and makes a meaningful contribution to reading comprehension theories. However, several limitations warrant consideration and should be addressed:

1. Introduction organization and theoretical foundation: The introduction requires restructuring to enhance coherence and establish a stronger theoretical foundation. Specifically, this research question has been previously investigated in two key studies that should be prominently cited and discussed as the foundational basis for the current work: (1) Huang, L., & Li, X. (2020). Early, but not overwhelming: The effect of prior context on segmenting overlapping ambiguous strings when reading Chinese. Quarterly Journal of Experimental Psychology, 73, 1382-1395; and (2) Huang, L., & Li, X. (2023). The effects of lexical- and sentence-level contextual cues on Chinese word segmentation. Psychonomic Bulletin & Review, 31, 293-302. These studies should be integrated into the introduction to provide the necessary theoretical groundwork and justify the current investigation's contribution to the existing literature.
2. The introduction should conclude with specific and coherent research hypotheses. The final section of the introduction lacks clearly articulated hypotheses that logically flow from the theoretical background and previous findings. The authors should formulate explicit, testable hypotheses regarding the expected relationships between contextual predictability and segmentation preferences, as well as potential interaction effects, to provide clear direction for the empirical investigation.
3. Regarding the methodology, the authors reported using G*Power to determine that 45 participants were needed. However, this calculation appears to assume a correlation among repeated measures of 0, which is unrealistic for eye-tracking studies. When recalculating with G*Power using a more appropriate correlation among repeated measures of 0.5, the required sample size would be 24 participants. Given that the authors ultimately recruited 76 participants, a clear justification is needed for this substantial increase beyond the calculated requirement. The authors should explain the rationale for selecting a larger sample size and discuss the specific advantages this provides, such as increased statistical power for detecting smaller effects, enhanced generalizability, or the ability to conduct additional analyses that were not initially planned.
4. When describing the recruitment of "Another group of 32 undergraduate students was recruited to evaluate the contextual predictability of the sentences," the authors should specify whether the contextual predictability ratings refer to the AB-C segmentation or the A-BC segmentation. This clarification is crucial for understanding how the predictability measure relates to the different segmentation possibilities and how it should be interpreted in the subsequent analyses.
5. The content in the third row of Table 2 is difficult to interpret and requires clarification. The authors should provide a clearer explanation or reformulation of this information to enhance readability and ensure that readers can properly understand the data being presented.

 

6. The current study found no significant interaction between contextual predictability and segmentation type, which contradicts findings from previous research, particularly "Predictability eliminates neighborhood effects during Chinese sentence reading," where contextual predictability was shown to modulate lexical neighborhood effects. I suspect that a major reason for this discrepancy lies in the inconsistent operational definitions of contextual predictability between the two studies, with the current study potentially having a smaller range of high versus low contextual predictability differences. The general discussion should address this limitation and acknowledge how the restricted range of predictability manipulation may have contributed to the null interaction effect. The authors should discuss how future studies might employ a broader range of contextual predictability to better capture its potential moderating effects on segmentation processes.
7. The authors' claim that the current findings support the Chinese Reading Model (CRM) and the graded pre-activation perspective appears to be an over-interpretationthat warrants further discussion. The strength of the evidence and the degree to which the results actually support these theoretical frameworks should be more carefully evaluated. The authors should provide a more nuanced discussion of how their findings relate to these models, acknowledging potential limitations in drawing such broad theoretical conclusions and considering alternative interpretations of their data.

Author Response

Response to Reviewer 1’s comments

Comments1: Introduction organization and theoretical foundation: The introduction requires restructuring to enhance coherence and establish a stronger theoretical foundation. Specifically, this research question has been previously investigated in two key studies that should be prominently cited and discussed as the foundational basis for the current work: (1) Huang, L., & Li, X. (2020). Early, but not overwhelming: The effect of prior context on segmenting overlapping ambiguous strings when reading Chinese. Quarterly Journal of Experimental Psychology, 73, 1382-1395; and (2) Huang, L., & Li, X. (2023). The effects of lexical- and sentence-level contextual cues on Chinese word segmentation. Psychonomic Bulletin & Review, 31, 293-302. These studies should be integrated into the introduction to provide the necessary theoretical groundwork and justify the current investigation's contribution to the existing literature.

Response1: We have added a discussion of the relevant literature by Huang and Li, as well as other related studies, in the introduction, specifically after line 114 on page 3 and after line 159 on page 4. We have also further strengthened the logical coherence among the studies and improved the fluency of the argumentation. In addition, we compared these studies with the results of the present study in the Discussion section, specifically after line 426 on page 11.

 

Comments2: The introduction should conclude with specific and coherent research hypotheses. The final section of the introduction lacks clearly articulated hypotheses that logically flow from the theoretical background and previous findings. The authors should formulate explicit, testable hypotheses regarding the expected relationships between contextual predictability and segmentation preferences, as well as potential interaction effects, to provide clear direction for the empirical investigation.

Response2: We have strengthened the summary of previous studies and the rationale for raising the research questions, and we have improved the formulation of the hypotheses. These revisions are presented after line 194 on page 5 of the manuscript.

 

Comments3: Regarding the methodology, the authors reported using G*Power to determine that 45 participants were needed. However, this calculation appears to assume a correlation among repeated measures of 0, which is unrealistic for eye-tracking studies. When recalculating with G*Power using a more appropriate correlation among repeated measures of 0.5, the required sample size would be 24 participants. Given that the authors ultimately recruited 76 participants, a clear justification is needed for this substantial increase beyond the calculated requirement. The authors should explain the rationale for selecting a larger sample size and discuss the specific advantages this provides, such as increased statistical power for detecting smaller effects, enhanced generalizability, or the ability to conduct additional analyses that were not initially planned.

Response3: We apologize for the incorrect parameter settings in G*Power, which led to an error in calculating the required sample size. We have recalculated the required number of participants and corrected it to 24, and we have provided a brief explanation for recruiting a larger sample, mainly regarding the improvement of ecological validity. These revisions can be found after line 211 on page 5 of the manuscript.

 

Comments4: When describing the recruitment of "Another group of 32 undergraduate students was recruited to evaluate the contextual predictability of the sentences," the authors should specify whether the contextual predictability ratings refer to the AB-C segmentation or the A-BC segmentation. This clarification is crucial for understanding how the predictability measure relates to the different segmentation possibilities and how it should be interpreted in the subsequent analyses.

Response4: The contextual predictability mentioned here refers to the degree to which the target word is predicted under different levels of contextual constraint—that is, the predictability of the two-character target word in sentences where the two-character segmentation is more plausible, or the predictability of the single-character target word when the single-character segmentation is more plausible. We have also added a brief description of this in the manuscript, as reflected in the revised description of the sentence evaluation on page 6.

Comments5: The content in the third row of Table 2 is difficult to interpret and requires clarification. The authors should provide a clearer explanation or reformulation of this information to enhance readability and ensure that readers can properly understand the data being presented.

Response5: We have also provided a further description of the evaluation measures in the sentence evaluation section on page 6, and we have revised the contents of Table 2 so that the order now corresponds to the sequence described in the manuscript, rather than the previously incorrect ordering.

Comments6: The current study found no significant interaction between contextual predictability and segmentation type, which contradicts findings from previous research, particularly "Predictability eliminates neighborhood effects during Chinese sentence reading," where contextual predictability was shown to modulate lexical neighborhood effects. I suspect that a major reason for this discrepancy lies in the inconsistent operational definitions of contextual predictability between the two studies, with the current study potentially having a smaller range of high versus low contextual predictability differences. The general discussion should address this limitation and acknowledge how the restricted range of predictability manipulation may have contributed to the null interaction effect. The authors should discuss how future studies might employ a broader range of contextual predictability to better capture its potential moderating effects on segmentation processes.

Response6: After reviewing the article you mentioned, we believe that its main research focus is not highly relevant to the topic of our study; therefore, we did not include it in the discussion or comparison of the present manuscript. However, following the direction suggested by your comment, we conducted additional comparisons between our findings and those of earlier studies in related areas. These revisions can be found in the Discussion section (from page 10).

Comments7: The authors' claim that the current findings support the Chinese Reading Model (CRM) and the graded pre-activation perspective appears to be an over-interpretationthat warrants further discussion. The strength of the evidence and the degree to which the results actually support these theoretical frameworks should be more carefully evaluated. The authors should provide a more nuanced discussion of how their findings relate to these models, acknowledging potential limitations in drawing such broad theoretical conclusions and considering alternative interpretations of their data.

Response7: We appreciate your comment pointing out that the Discussion section was not sufficiently developed. Based on the current results and the models supported by our findings, we have provided further discussion and elaboration, which can also be found in the revised Discussion section (from page 10).

All modification are shown in the attachment. Thank you!

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript reports an eye-tracking experiments to investigate the impact of contextual predictability on word segmentation during Chinese reading. The results showed that contextual predictability significantly affected skipping probability but had no effect on fixation durations. Additionally, the AB-C segmentation showed a general advantage compared to the A-BC segmentation, which was observed across several fixation duration measures. This paper explores an interesting topic. However, there are several concerns that need clarification.

1. The authors mixes “overlapping ambiguous words” and “overlapping ambiguous strings” throughout the manuscript. However, it is worth noting that overlapping ambiguous strings are not words and most of them cannot be found in dictionaries. Instead, they are three-character strings that contain word boundary ambiguity. And the authors use the English trimorphemic word “unlockable” as an analogy on page 2. However, there is an intrinsic difference between English trimorphemic word and overlapping ambiguous strings, and the processing mechanisms involved may differ. This point should be clarified.

2. The research questions and motivations of the study need further clarification.

1) The authors introduce studies and theories suggesting that Chinese readers have a preference for two-character words. Given that this finding is relatively well established, the motivation for the present study needs to be clarified.

2) A review of the literature on the influence of contextual predictability on word processing in Chinese reading is lacking. The first paragraph on page 3 introduces the impact of context on word processing during English reading, rather than supporting the claim that “contextual predictability is shown to influence lexical processing during Chinese reading” in line 106 on page 3. After supplementing relevant literature on the predictability effects in Chinese word processing, the authors also need to clarify the motivation of investigating the predictability effects on Chinese word segmentation.

3) Two hypotheses are proposed on page 4, with the second hypothesis assuming that contextual predictability operates independently of the word segmentation process. Is there theoretical support for this assumption?

3. The manipulation and evaluation of contextual predictability should be clarified. Based on Table 2, it appears that the predictability in this study was measured using a 7-point scale. Why was cloze probability (the classic measure of predictability) not used? What were the instructions? Was the manipulation based on predictions of the overlapping ambiguous strings, or predictions of word A/word AB? According to the results and discussion, the authors intended to manipulate the contextual predictability of overlapping ambiguous string. However, an overlapping ambiguous string is not a word but a three-character string, and it is difficult to for participants to predict a three-character string. Moreover, based on Table 2, the value of 4.17 in the low predictability condition indicates relatively low predictability rather than absolute low predictability.

4. Minor issues

- Are there references for lines 109-112 on page 3?

- There is a typo in the Chinese example on page 4, line 155.

- The Method section should report how many sets of stimuli were included.

- In Table 1, the meaning of the numerical values for “segmentation pattern” should be clarified.

- On page 7, the results for the main effect are not clear and the specific pattern (e.g., which condition was higher or lower) should be explicitly stated.

- In Figure 2, the plot appears to show an interaction pattern rather than a main effect for the skipping rate, and the y-axis label contains an error.

Author Response

Comments1: The authors mixes “overlapping ambiguous words” and “overlapping ambiguous strings” throughout the manuscript. However, it is worth noting that overlapping ambiguous strings are not words and most of them cannot be found in dictionaries. Instead, they are three-character strings that contain word boundary ambiguity. And the authors use the English trimorphemic word “unlockable” as an analogy on page 2. However, there is an intrinsic difference between English trimorphemic word and overlapping ambiguous strings, and the processing mechanisms involved may differ. This point should be clarified.

Response1: We have re-distinguished and discussed the concepts of overlapping ambiguous words and overlapping ambiguous strings, and, drawing on previous research, we have explained the rationale and justification for using overlapping ambiguous strings in the present study. See after line 114 on page 3 and after line 166 on page 4.

 

Comments2: The research questions and motivations of the study need further clarification.

1) The authors introduce studies and theories suggesting that Chinese readers have a preference for two-character words. Given that this finding is relatively well established, the motivation for the present study needs to be clarified.

2) A review of the literature on the influence of contextual predictability on word processing in Chinese reading is lacking. The first paragraph on page 3 introduces the impact of context on word processing during English reading, rather than supporting the claim that “contextual predictability is shown to influence lexical processing during Chinese reading” in line 106 on page 3. After supplementing relevant literature on the predictability effects in Chinese word processing, the authors also need to clarify the motivation of investigating the predictability effects on Chinese word segmentation.

3) Two hypotheses are proposed on page 4, with the second hypothesis assuming that contextual predictability operates independently of the word segmentation process. Is there theoretical support for this assumption?

Response2: The aim of the present study is to discuss readers’ preferences in sentence reading under different contextual conditions, in cases where both two-character words and single-character words can form reasonable sentences, rather than focusing solely on preferences for two-character words. The more frequent use of two-character words is taken only as background for the study and as one of the hypotheses to be examined. Subsequently, we supplemented and discussed relevant research in Chinese reading, and further refined and revised the research hypotheses. See line 82 on page 2, line 115 on page 3, line 159 on page 4, and line 196 on page 5.

 

Comments3: The manipulation and evaluation of contextual predictability should be clarified. Based on Table 2, it appears that the predictability in this study was measured using a 7-point scale. Why was cloze probability (the classic measure of predictability) not used? What were the instructions? Was the manipulation based on predictions of the overlapping ambiguous strings, or predictions of word A/word AB? According to the results and discussion, the authors intended to manipulate the contextual predictability of overlapping ambiguous string. However, an overlapping ambiguous string is not a word but a three-character string, and it is difficult to for participants to predict a three-character string. Moreover, based on Table 2, the value of 4.17 in the low predictability condition indicates relatively low predictability rather than absolute low predictability.

Response3: We have provided an explanation of the method used for rating the sentences. In addition, the aim of the present study is to examine participants’ reading and word-segmentation performance under high and low contextual predictability conditions. Therefore, in contrast to previous studies that used neutral contexts, the contexts used in the present study were high-predictability and low-predictability contexts. See after line 239 on page 6.

 

Comments4:  Minor issues

- Are there references for lines 109-112 on page 3?

- There is a typo in the Chinese example on page 4, line 155.

- The Method section should report how many sets of stimuli were included.

- In Table 1, the meaning of the numerical values for “segmentation pattern” should be clarified.

- On page 7, the results for the main effect are not clear and the specific pattern (e.g., which condition was higher or lower) should be explicitly stated.

- In Figure 2, the plot appears to show an interaction pattern rather than a main effect for the skipping rate, and the y-axis label contains an error.

Response4: All minor issues have been revised, as shown at line 110 on page 3, line 220 on page 5, Table 1 on page 6, lines 319 and 326 on page 8, and in Figure 2. Regarding Figure 2, this figure illustrates that the differences in predictability vary across the two segmentation conditions, but it does not address the differences between the two segmentation conditions themselves. Therefore, we consider that this figure depicts the main effect of predictability.

All modification are shown in the attachment. Thank you!

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have made lots of revisions based on thoughtful thinking, but several major issues remain and should be addressed.

 

Abstract:

1. In line 17, the term“overlapping ambiguous words” is still used and should be corrected. In line 21, “target words” should be revised to “target strings/region”. Similar terminological errors appear throughout the manuscript (e.g., Page 4, Line 177; Page 5, Line214, Line 226; Page 7, Line 289; P10, Line 344...). The authors should carefully check and correct these inconsistencies across the entire manuscript.
2. Inline 18, the authors should clarify the contextual predictability was manipulated for what. And the segmentation type “AB-C” and “A-BC” should be defined.

 

Introduction:

3. The second paragraph on Page 2 discussesthe role of inter-word space in Chinese reading. It is unclear how this paragraph is related to the present research question.
4. On Page3, Line 116, the “AB-C” segmentation type is mentioned without explanation. And “overlapping ambiguous string” has not yet been clearly defined.
5. On Page 4, Line 165, the statement that“Chinese reading involves a large number of overlapping ambiguous strings”, would be clearer if authors provide the proportion information.
6. On Page 4,Line 180, there is a typo “跟” in the Chinese example that should be corrected.
7. One Page 4, Line 170, “readers’ segmentation preferences may stem from the requirement for semantic coherence rather than from contextual predictability itself.” and Line 182 “the influence of semantic information can be ruled out, allowing for an isolated examination of how contextual information affects word segmentation.” It is confusing and it is hard to understand these statements. Do the authors mean that the present study can rule out the effect of “semantic integration”?
8. Page 4-5, in the hypotheses part, the authors present only one hypothesis, and competing alternative hypotheses are not clearly stated, nor are specific predictions based on different hypotheses.

 

Method:

9. In Table 1, the meaning of the numerical values for “segmentation pattern” is still not clear. For example, what does the value “20.84” represent?
10. Only predictability ratings for high vs. low predictability conditions are reported. But as shown in the examples, the prior context differs across the four conditions. It is unclear why predictability ratings were collapsed across AB-C and A-BC conditions. Thus, the predictability manipulation remains confusing and insufficiently justified.

 

Results:

11. The authors should report the details of data analysis. Were linear mixed-effects models used? If so, the details of models should be clearly described, including fixed effect,s random effects, and whether random slopes were included.
12. The authors should clarify how Bayes factor are interpreted when values are lower than 1.

 

Discussion:

13. The explanation for the null effect of interaction between contextual predictability and word segmentation (Page 12) is  not fully convincing. One important limitation of the study is that the predictability was manipulated at the level of strings with different segmentation type (AB-C vs. A-BC). In contrast, predictability is typically defined at the level of a word (e.g., AB or A), making it difficult for contextual predictability to directly influence string-level segmentation structure. In addition, as acknowledged on Page 11, the “low predictability” condition is not truly low: the predictability rating is higher than 4 on a 7-point scale. Although the difference between low and high predictability conditions is statistically significant, both conditions still impose substantial constraints on following continuation. This may have limited the possibility of observing an interaction effect.

Author Response

Comments1: In line 17, the term“overlapping ambiguous words” is still used and should be corrected. In line 21, “target words” should be revised to “target strings/region”. Similar terminological errors appear throughout the manuscript (e.g., Page 4, Line 177; Page 5, Line214, Line 226; Page 7, Line 289; P10, Line 344...). The authors should carefully check and correct these inconsistencies across the entire manuscript.

Response1: We have further reviewed and revised the use of terminology in the manuscript and have indicated the corresponding changes in the text.

 

Comments2: Inline 18, the authors should clarify the contextual predictability was manipulated for what. And the segmentation type “AB-C” and “A-BC” should be defined

Response2: We have clarified the purpose of the predictability manipulation in line 15, namely to test the experimental hypothesis, and provided examples of the word segmentation conditions in line 18.

 

Comments3: The second paragraph on Page 2 discusses the role of inter-word space in Chinese reading. It is unclear how this paragraph is related to the present research question.

Response3: We have revised this paragraph and added a brief clarification of its connection with the subsequent discussion; see the text following line 70.

 

Comments4: On Page3, Line 116, the “AB-C” segmentation type is mentioned without explanation. And “overlapping ambiguous string” has not yet been clearly defined.

Response4: We have provided examples of overlapping ambiguous strings and elaborated on the concept in the manuscript (see line 115).

 

Comments5: On Page 4, Line 165, the statement that“Chinese reading involves a large number of overlapping ambiguous strings”, would be clearer if authors provide the proportion information.

Response5: We have cited a relevant study discussing the prevalence of overlapping ambiguous words in Chinese (see line 172).

 

Comments6: On Page 4,Line 180, there is a typo “跟” in the Chinese example that should be corrected.

Response6: We have corrected this spelling error.

 

Comments7: One Page 4, Line 170, “readers’ segmentation preferences may stem from the requirement for semantic coherence rather than from contextual predictability itself.” and Line 182 “the influence of semantic information can be ruled out, allowing for an isolated examination of how contextual information affects word segmentation.” It is confusing and it is hard to understand these statements. Do the authors mean that the present study can rule out the effect of “semantic integration”?

Response7: This paragraph is intended to highlight that previous studies have reported mixed findings regarding the roles of context and semantics in reading, with some research suggesting that semantic information, rather than contextual information, exerts a stronger influence on word segmentation during reading. Accordingly, the present study minimized the influence of semantic factors through experimental manipulation in order to specifically examine the effect of contextual information on word segmentation.

 

Comments8: Page 4-5, in the hypotheses part, the authors present only one hypothesis, and competing alternative hypotheses are not clearly stated, nor are specific predictions based on different hypotheses.

Response8: We proposed three research hypotheses and provided additional clarification of their content in the manuscript (see line 206).

 

Comments9: In Table 1, the meaning of the numerical values for “segmentation pattern” is still not clear. For example, what does the value “20.84” represent?

Response9: We have added a more detailed description of this term starting from line 239.

 

Comments10: Only predictability ratings for high vs. low predictability conditions are reported. But as shown in the examples, the prior context differs across the four conditions. It is unclear why predictability ratings were collapsed across AB-C and A-BC conditions. Thus, the predictability manipulation remains confusing and insufficiently justified.

Response10: We have treated the evaluation of word segmentation and the evaluation of context separately as segmentation accuracy and contextual predictability, respectively, and have provided explicit definitions and discussion of these two terms in the manuscript (starting from line 253).

 

 

Comments11: The authors should report the details of data analysis. Were linear mixed-effects models used? If so, the details of models should be clearly described, including fixed effect,s random effects, and whether random slopes were included.

Response11: We have discussed the measures you referred to starting from line 315.

 

Comments12: The authors should clarify how Bayes factor are interpreted when values are lower than 1.

Response12: We have provided a more detailed explanation of the Bayesian results from line 364.

 

Comments13: The explanation for the null effect of interaction between contextual predictability and word segmentation (Page 12) is  not fully convincing. One important limitation of the study is that the predictability was manipulated at the level of strings with different segmentation type (AB-C vs. A-BC). In contrast, predictability is typically defined at the level of a word (e.g., AB or A), making it difficult for contextual predictability to directly influence string-level segmentation structure. In addition, as acknowledged on Page 11, the “low predictability” condition is not truly low: the predictability rating is higher than 4 on a 7-point scale. Although the difference between low and high predictability conditions is statistically significant, both conditions still impose substantial constraints on following continuation. This may have limited the possibility of observing an interaction effect.

Response13: Thank you for your helpful suggestion. We have made minor revisions to overly absolute statements in the Discussion (see lines 451 and 485). The present study builds on the work of Huang and Li (2020, 2023), which demonstrated that contextual information influences the segmentation of ambiguous strings. However, in those studies, contextual information was treated in a relatively categorical manner, distinguishing only between the presence and absence of context. Extending this line of research, our study further examined the role of contextual information in Chinese word segmentation during reading, with particular attention to the degree of contextual constraint. Our results showed that, even under high- and low-predictability conditions, no significant interaction between contextual predictability and word segmentation was observed, and the absence of such an interaction received moderate support from the Bayesian analyses, indicating the plausibility of our findings. We also plan to further control and investigate the issues you raised in future research.

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

I thank the authors for the responses and have only a few remaining concerns about the Discussion and some inaccurate citations.

 

1. Are there references for lines 67-68 on Page 2?

 

2. Page 4 lines 158-161. Huang and Li (2023) is cited incorrectly. The study does not examine the effect of contextual predictability.

 

3. Page 5, segmentation pattern preference. What I would like to know is the interpretation of the reported values. For example, does 20.84 denote that 20.84 out of 32 (total participants) preferred to perform AB-C segmentation?

 

2. Page 10. It would be clearer if the authors explicitly stated the logic behind their claim that contextual predictability represents an active top-down process, whereas word frequency reflects bottom-up passive input, and how this distinction accounts for the present null effects on fixation durations, given that previous studies have found significant effects of contextual predictability on fixation durations.

 

3. Page 12 lines 485-487. The citations are incorrect. These two studies do not claim that contextual information has little impact on word segmentation.

 

4. Page 12 lines 489-494. The authors state that readers “…activate potential candidate words in parallel, completing segmentation by activating the candidates with higher predictability.” Here, the use of “predictability” appears to be conflicting. Because the authors later state that “This process relies primarily on the internal structural properties of the lexical string and the activation levels of neighboring words, RATHER THAN on the direct modulation of contextual predictability,” which seems to suggest that word segmentation is unaffected by predictability.

 

5. Page 13. The original paper of CRM (as mentioned in Reference list - An Integrated Model of Word Processing and Eye-Movement Control During Chinese Reading) states that “The influence of word predictability is implemented as increasing the activation of the corresponding word unit.” If my understanding is correct, this suggests that predictability may affect competition by influencing the activation level of word units. This interpretation is inconsistent with the statement in lines 501-504 that “the CRM model’s prediction that contextual information…is insufficient to change the competition among candidate target character strings.”

 

6. It would be helpful to carefully check details throughout the manuscript. For example, the titles of tables and figures require revision, the figure could be clearer, and some errors appear on Page 4 (line 166-168), Page 5 (line 229), Page 7 (line 295) and Page 5 (line 296).

Author Response

Comments1: Are there references for lines 67-68 on Page 2?

Response1: We have added the citation at line 68 in the original text.

 

Comments2: Page 4 lines 158-161. Huang and Li (2023) is cited incorrectly. The study does not examine the effect of contextual predictability.

Response2: We carefully reread the article and found that it primarily discusses the influence of word-level information rather than contextual factors. We apologize for our earlier misinterpretation of its content, and we have therefore removed this citation from the relevant section.

 

Comments3: Page 5, segmentation pattern preference. What I would like to know is the interpretation of the reported values. For example, does 20.84 denote that 20.84 out of 32 (total participants) preferred to perform AB-C segmentation?

Response3: Your understanding is correct. In addition, to clarify why the total number exceeds 32, we have added further explanation at line 232 of the original text to facilitate understanding.

 

Comments4: Page 10. It would be clearer if the authors explicitly stated the logic behind their claim that contextual predictability represents an active top-down process, whereas word frequency reflects bottom-up passive input, and how this distinction accounts for the present null effects on fixation durations, given that previous studies have found significant effects of contextual predictability on fixation durations.

Response4: We have added a supplementary explanation at line 387 to facilitate readers’ understanding.

 

Comments5: Page 12 lines 485-487. The citations are incorrect. These two studies do not claim that contextual information has little impact on word segmentation.

Response5: Thank you for your reminder. We have made minor revisions to this summary starting at line 484 and reorganized the citations accordingly.

 

Comments6:Page 12 lines 489-494. The authors state that readers “…activate potential candidate words in parallel, completing segmentation by activating the candidates with higher predictability.” Here, the use of “predictability” appears to be conflicting. Because the authors later state that “This process relies primarily on the internal structural properties of the lexical string and the activation levels of neighboring words, RATHER THAN on the direct modulation of contextual predictability,” which seems to suggest that word segmentation is unaffected by predictability.

Response6: We apologize that the wording conflicted with the intended meaning. Starting from line 492, we have revised this passage to better align it with the results of our study.

 

Comments7: Page 13. The original paper of CRM (as mentioned in Reference list - An Integrated Model of Word Processing and Eye-Movement Control During Chinese Reading) states that “The influence of word predictability is implemented as increasing the activation of the corresponding word unit.” If my understanding is correct, this suggests that predictability may affect competition by influencing the activation level of word units. This interpretation is inconsistent with the statement in lines 501-504 that “the CRM model’s prediction that contextual information…is insufficient to change the competition among candidate target character strings.”

Response7: In fact, what we intended to convey is that the influence of context is largely grounded in the intrinsic properties of lexical items. For example, a preference for two-character words does not shift to a preference for single-character words simply as a result of contextual influence. Nevertheless, our original wording led to misunderstanding. Therefore, starting from line 502, we have revised the expression to minimize potential ambiguity in the manuscript.

 

Comments8: It would be helpful to carefully check details throughout the manuscript. For example, the titles of tables and figures require revision, the figure could be clearer, and some errors appear on Page 4 (line 166-168), Page 5 (line 229), Page 7 (line 295) and Page 5 (line 296).

Response8: We have asked other members of our team to conduct an additional review of the manuscript and have revised it as thoroughly as possible to address issues such as incorrect terminology, spelling errors, and formatting problems. In particular, we have clearly marked the points you mentioned in the manuscript. Thank you very much for your reminder and careful review.