Regular and Irregular Inflection in Different Groups of Bilingual Children and the Role of Verbal Short-Term and Verbal Working Memory

Round 1

Reviewer 1 Report

The article is well-written, well-organized and represents a good contribution to research about the acquisition of inflection in bilingual children. In particular, it deals with a longitudinal study about noun plural and past participle formation in Dutch by bilingual children, who have different weak languages (Arabic, Frisian, Tarifit, Turkish).

The authors’ aim is to investigate the role of regularity, cross-linguistic distance/proximity, verbal short-term (VSTM) and working memory (VWM) in bilingual performance. They also take into account the impact of amount of Dutch (input) and parental education as variables.

For the study, 231 five-to-eight-year old children were analyzed (45 Dutch monolingual, 106 Frisian-Dutch bilinguals, 31 Turkish-Dutch bilinguals, 38 Tarifit-Dutch bilingual and 10 Arabic-Dutch bilinguals) and even though the groups are of unequal sizes, data analysis is clearly exposed via statistical procedures. The tasks used in the research are part of a battery including language, working memory and attention tasks.

Results show that an important use of Dutch at home, an older age and a higher level of parental education may be associated with a smaller gap between accuracy at using regular and irregular inflection. On the contrary, the hypothesised effect of the cross-linguistic similarity between the dominant language and the weak one does not play a major role in the bilingual performance. For instance, Frisian is typologically close to Dutch as for lexical and grammatical overlaps. Nevertheless, the study shows that Frisian-Dutch bilingual children do not exploit this proximity between the two languages. Moreover, a binary categorical distinction between monolingual and bilingual groups seem not to be possible: differences can also be found within the bilingual groups. Eventually, children with better VSTM, but not VWM, were more accurate at using regular and irregular inflection. This demonstrates that when VSTM, VWM and non-verbal intelligence are all taken into account, non-verbal intelligence is more important than VWM.

The theoretical background of the article is really strong, but I suggest the authors to add some references for some notions (minority language vs. majority language; positive transfer vs. negative transfer) as I noted at pages 3 and 5. I also put some comments on the article that the authors could take into account in order for the article to be more exhaustive for the readers.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Brief summary: As bilingual children often manifest problems with inflectional morphology, this longitudinal study aim (i) to examine how different bilingual groups perform on regular and irregular noun plurals and past participles and (ii) to investigate how this performance is related to verbal short-term memory (VSTM) and verbal working memory (VWM). Measures from 231 typically developing children (5-8 years age) were collected. All participants were learners of Dutch who have either Arabic, Frisian, Tarifit, or Turkish as their second language.

Broad comments: The strength of this study is the analysis of the relationship between linguistic variables and VSTM / VWM. However, my main concerns are: (i) to some extent the paper lacks a clear rationale and hypotheses, (ii) the paper lacks newsworthiness, and (iii) the study just applied single global measures of VSTM and VWM. 

Specific comments:

Abstract

To give a quick insight into the overall approach and procedures, research methods should be inserted in the abstract.

On page 1, line 21, the authors wrote “showing that a binary categorical distinction between monolingual and bilingual groups is untenable”, however, research has repeatedly shown that bilingualism is not a categorical variable (e.g., Gigi Luk & Ellen Bialystok, 2013). Thus, in my opinion, the paper lacks newsworthiness.

Introduction

It seems that the authors first decided on important methodological issues and only later found the theoretical explanations (e.g., choice of these specific bilingual groups; page 3, line 111). In my opinion, the introduction should first set the theoretical background and point out the gap in knowledge that the rest of the paper will fill. Thus, I suggest start broadly and then narrow down. It is important to understand the theoretical factors that motivate the methodological choices. Some questions that the introduction should answer clearly: 

• Why these minority languages?
• Why VSTM and VWM? Other studies have found relations with other executive functions such as inhibition (e.g., Bialystok et al., 2004; Costa et al., 2009; Pelham and Abrams, 2014) and task-switching (e.g., Prior and Gollan, 2011; Wiseheart et al., 2016).
• Why the authors design a longitudinal study?
• What about non-verbal intelligence? This variable just appears in the method. Why the authors assess this skill?

I also suggest that the authors present their specific hypotheses in the section “present study” and report the results accordingly.

Method

- Why did the authors not assess other factors relevant to variation in bilingualism as the age of L2 acquisition and years of residence in the country?

- Page 5, line 244: the authors wrote: “All children were 231 typically developing with no indication of language and speech disorders.” How did the authors evaluate this? 

- In particular, for the statistical analysis conducted and for group comparison, the authors should have an equivalent number of participants per group. For instance, the Frisian group has 106 participants, whereas the monolinguals group has 45, and the Arabic 11. Also, the gender distribution is not equivalent for the Arabic group. 

- Some measures are just presented in the participants’ section, although they are also presented in the results section. I think that these measures should also appear in the measures section. 

- In my opinion, the use of codes to represent variables in the text, tables, and graphs is confusing (e.g., Digit_fw, Digit_bw, Lemma_freq, EdPar…).

Results

- Please be careful with the conventions for figures and tables, namely with titles, legends explaining symbols, and vertical / horizontal tick marks. For instance, in Figure 8 the title is “Predicted probabilities of correct” and a more adjusted title should be “Predicted probabilities of correct responses for the interaction between @ and @”. 

- A major limitation of this study was the failure to use more than one indicator per construct (i.e., verbal short term memory and working memory). This indeed makes the findings very difficult to interpret. If the authors had used three or more indicators per construct, then they could have explored the role of verbal short term memory and working memory. This is a serious structural problem of their study. For instance, Kroll and Bialystok studied the consequences of bilingualism for language processing and cognition, and they suggest that reducing performance to a few measurable components fails to capture the most crucial outcomes. So, I think that the authors should report this limitation and be careful with statements such as: “Analytical reasoning (indexed by non-verbal intelligence) is more important than VWM for children’s accurate use of inflection.“ (Page 21, line 707) because working memory was assessed with a single task.

- The article has too many tables and graphs. For instance, tables in Appendix 2 can be presented in just one table: correlations above the diagonal for all children and below the diagonal for bilingual children. 

- All Graphs and Tables should also be referred to within the text.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The manuscript has been significantly improved. However, the following minor comments were not properly addressed:
Abstract: The methods section should contain enough information to enable the reader to understand what was done and how. It is usually the second-longest section in the abstract. In my opinion, the authors should reformulate the abstract. 
The authors referred that they have already presented their specific hypotheses. However, in the section “present study” they didn't present hypotheses. This is an example of a hypothesis: "Students who experience test anxiety before an English exam will get higher scores than students who do not experience test anxiety”.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf