Effects of Input Consistency on Children’s Cross-Situational Statistical Learning of Words and Morphophonological Rules

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

MS: languages-3306253

Title: Effects of Input Consistency on Children’s Cross-Situational Statistical Learning of Words and Morphophonological Rules

 

This paper aimed to investigate how errors in linguistic input affect language learning in 7-11 year old children. They found that whilst children learned the words, there was no evidence that they had learned the morphophonological rules even in a 100% correct input condition. Further that errors in the input led to decreased learning of words. The authors aim for this to contribute to understanding on how differences in environments affect language learning.

 

Overall, this paper was clear, concise, and well written. The study design is clear and interesting. However, I unfortunately have some concerns about the validity and ultimate use of the data. Some of these may well become less relevant with some explanation on some aspects below. There is a wide age range presented within relatively small group sample sizes and vast gender differences and the data analysis could do more to investigate whether these are relevant factors. It is usually quite standard to include some sort of language ability questionnaire to ensure this doesn’t explain results across groups but this hasn’t been included here. More pressing though is that the authors report a large number of learning studies that were conducted at the same time. That, and the lack of learning in the control, consistent condition, leads me to worry that there is something happening with attention in this study. This may also explain the animacy effect whereby children might be managing to hold attention to animate object but not others. Further, that even if the data were more clear in their validity, I think it may offer a more pessimistic view of error variability that necessary. The authors note that this type of error is quite rare and, to add to this, children of this age often question to reduce ambiguity and statistical learning is now one of many language learning strategies.

 

Abstract and introduction

1.     Both abstract and introduction are clear and literature nicely is described. The aims and hypotheses are clear. You might want to read and think about if relevant to include the following paper on the subject of input variability.

Dunn, K. J., Frost, R. L., & Monaghan, P. (2024). Infants’ attention during cross-situational word learning: Environmental variability promotes novelty preference. Journal of Experimental Child Psychology, 241, 105859.

2.     There is a section related to differences in generalisability and grammatical ability in children – including this in detail led me to think you’d be controlling for this or investigating this. Given the wide age range it seems like it would be important to factor in?

 

Methods.

 

3.     Overall the methods and procedures were very clear and nicely illustrated.

4.     Looking at the sample characteristics, there is quite a vast gender imbalance in the consistent and inconsistent groups. Have you investigated any differences in the data between girls and boys? Girls can show more advanced language ability in some samples at this age and it might explain your reversal effect of animacy between consistent and inconsistent groups.

5.     Again on sample characteristics – do you have a justification for sample size with such a wide age range? Typically, data can see effects in these studies with this size per condition but with much narrower age ranges.

6.     Overall the methods and procedures were very clear and nicely illustrated.

7.     Given that there is no learning of the morphophonological rules and that this age range is capable of doing so – there might have been something about the task that was too complex? You mention lots of other language learning tasks in lines 344 to 351 – were these presented in the same order each time and where did this study fall? Could you have saturation or fatigue effects? This does seem like a lot of learning studies in this time, particularly for the younger children in the sample.

 

Results

8.     Really clear analysis and reporting

9.     Why were inconsistent conditions combined for analysis and was there a correction in place for the double of the sample size? Could the stats be reported for the comparison of consistent with each inconsistent condition separately.

 

Discussion

 

10.  Aside from the first line (497-498) where it looks like there is a slight misrepresentation if I understand it correctly (that inconsistent input hindered rule learning. Rule learning was poor even in consistent condition so it’s not clear what has led to this), the discussion does a good job or reporting results in context without overinterpretation. So much so that I found myself wondering why some of the methodological choices had been made in light of the evidence that is reported.

11.  However, on the subject of the limited application of these results, I would also add that any errors through statistical learning paradigms alone at this age offer limited applicability unless you assume that this is the only learning method. Children of 7-11 all have the ability (and use it) to ask questions when unclear, and corrections are often made. So any negative effects might offer a pessimistic view.

 

 

Author Response

Please see the attachment. 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper offers an analysis of the effects of (in)consistent input on 7-11-years-old Dutch-speaking children's cross-situational statistical learning of words and morphophonological rules in an artificial language. Therefore, the study has the potential to fill this gap in the field. However, I recommend addressing several issues:

 

1.             Abstract:

-I suggest using another word than "assess" since it is repeated twice in two lines (6 and 9).

 -Please, specify which differences are expected to be found in the quality of children's language environments that can affect their language learning.        

 

2.             Keywords:

I suggest including keywords related to the L1 Dutch and the artificial language of the present study.

 

3.             Introduction:

-The literature review is important and interesting to the present study; however, I recommend avoiding hypotheses and possible conclusions in this section. For instance, the information located on lines 42-50 should appear at the end of the introduction.

-Line 51: I recommend changing the words "studies" and "to study", given that it sounds repetitive.

- Line 87: Please, specify the year of the Yu and Smith's study. 2007?

-After line 165, I suggest adding information about how the paper is structured (how many sections there are, and in which consists each of them).

 

4.             Terminology:

-It seems a little bit unclear how the following terms are used in this paper: ‘artificial language’, 'statistical learning', ‘inconsistent input' and 'characteristics of the language input' as well as the possible effects of input. I suggest explaining better these concepts. For example, why the artificial language is important to assess this type of acquisition? What means inconsistent input? Does it refer only to substitution errors? On which characteristics of the language input are focused the present study? Quality/quantity factors? Which possible effects are reviewed?

 

5.             The present study:

-I suggest adding this part of the article in a separated section where the collected data and the procedure are explained.  

-This section presents repetitive information. First, the lines 168-170 seem to present the same information in comparison to the lines 177-178. Second, lines 170-174 present the same information than lines 46-50.

-Lines 204-211: Does this study analyses factors such as linguistic proficiency, age, multilingualism, socio-economic status and so on? If so, these should be included and reviewed before the present study.

 

6.             Research questions and hypotheses:

-I recommend adding this section after explaining the present study section.

 

 

7.             Materials and methods:

-I suggest including the information of the participants and materials before explaining the research questions and hypotheses.

 

8.             Participants:

-Are the participants Dutch monolingual speakers? How was it decided which participant received consistent or inconsistent input? Why the number of participants is so different across gender and input conditions? (see table 1).

-Is the maternal education level correlated with the children's results? Why is it expected this information is an important factor?

 

9.             Materials:

-Section 2.2 about the material is very clarifying. The explanation of the different blocks of the task is concise and illuminating.

-Lines 261-262. Please provide information on the total number of trials responded per child.

-Is there any relationship between the suffixes used in the artificial language of this study with some Dutch suffixes?

 

10.          Data analysis:

-Lines 362-368: The research question 1 is not divided into 1a-1b. If it is the aim, I suggest specifying that in the corresponding section.

 

11.          Results:

-The results section is so clear and interesting. Congratulations!

 

12.          Limitations and future directions:

-If limitations are found, how will they be addressed? In what direction will future research be directed?

 

13.          Notes:

-Note 2: Does it mean that all children participating in this study have Dutch-monolingual parents?

 

 

 

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf