Multisensory Interactive Digital Text for English Phonics Instruction with Bilingual Beginning Readers

Round 1

Reviewer 1 Report

In the acquisition of an alphabetic language, previous studies showed that teaching letter-sound correspondence through the picture mnemonic technique, i.e., integrating a picture cue with a letter form in which the initial sound of the name of the pictured object resembles the letter sound, was found to be effective in improving language learners’ early literacy skills (e.g, Ehri, 2014). In the current study, the authors employed a digital tool, iPad-based reading program, to showing a dynamic auditory-visual presentation of English letter-sound correspondence to bilingual children. Through a long-term (14 weeks) extensive training to 90 children in kindergarten (48 out of them in the experimental reading condition and 42 in the control condition), it was found that bilingual children benefited from this technology-based reading intervention and performed better in the measures of decoding at the post-test and word reading at both the post- and follow-up tests, but not in the measure of spelling. The authors further reported that letters with multiple sound mappings or specific sounds to the English language were challenging to acquire for the early language learners.

In this paper, there are several strengths. The studies related to the effect of the interactive digital teaching tool on early bilingual learners are relatively scarce. Additionally, this study involves with an investigation of teaching English letter-sound pairings to bilingual language learners who speak two typologically distant languages (i.e., Chinese and English). This bilingual group engage with different phonological awareness at the level of phonemes, which is different from the previous studies majorly focused on monolinguals in the alphabetic language context. This represents an important addition to the literature. 

There are just a few limitations, mostly related to data analysis and discussion that I detail as follows.

1. On page 12, line 490, there exist some problems in Figure 3 concerning the results of Word Reading. The authors collected data of the dependent measures in word reading, decoding, and spelling across 3 time points. However, the relevant graph of word reading shows 4 time-point intervals on the x-axis, which is inconsistent with the author’s previous description in the previous statement “Time (pretest, posttest, and follow-up) was entered as a repeated measures factor, ….” in line 480 on page 11. 

I noticed that Word Reading was also assessed as a baseline measure, which may account for the presence of 4 time points (baseline, pre-test, post-test, follow-up) exclusively for Word Reading. Actually, the authors have reported the result of baseline measures on page 10 lines 460-462, “One-way ANOVA’s showed that the groups (experimental and control) did not differ on any of these measures prior to the intervention (all p’s > 0.61) ”.  So, my suggestion is that to keep reporting 3 time points in this graph, which is consistent with the other two measures (decoding and spelling).  Additionally, there was an absence of explicit indicators on the x-axis to show the levels of the time-point variable in this graph of Word Reading. 

For all 3 graphs in Figure 3, the labels for the y-axis were missing. What is the scale for y-axis, e.g. the number of words? These relevant details should be provided in the graphs.

Please review Figure 3, use the relevant graph, and put the clear indicators in the graphs.

2. On page 12, lines 495-499, please add a reference to justifying the calculations of effect sizes. Moreover, clarity is lacking in these statements. The authors stated “subtracting the gain score of the control group from that of the reading group”. To improve clarity, it should be “subtracting the mean score of the control group from ….”. Importantly, I assume that the authors employed the Cohen’s d effect sizes. The common interpretation is to use the hierarchical degrees to refer to the effect sizes, such as small, medium or large. Therefore, it is inappropriate to use the expression of “not significant” in line 499 on page 12 to report the results. Similar inappropriate expressions were also present on page 16, lines 579-581. Please re-interpret the results in proper words.

3. On page 13, line 511, again, in the text of the manuscript, the authors stated that “For the story search task, search time was the average time to find all instances of a target grapheme.” However, in Figure 5, the authors used “Median Time per Grapheme (sec)” and “Median Time per letter” on the y-axis to indicate children’s performance in the story search and tracing tasks. In the statistical analysis, the terms Median (a middle point of a number set) and Mean (the average value of a number set) have totally different connotations. So, why is there a difference between the in-text statement and the indication in the figure, and which term is the correct one to be used in your analysis here? 

4. On page 13, lines 508-512, the authors mentioned that they used time responses to evaluate children’s performance in the tracing and story search tasks, while the measure of accuracy rates was adopted to report results in the spelling task. Why did the authors selectively use accuracy rates for one of the three measures? Why was this choice made? Accuracy and response times are the two dimensions involved in the experimental studies. The former one is related to the choice probability or how easily the decision can be made, while the latter is focused on the speed with the decisions made. Therefore, researchers are also interested in reading the results of response times in this spelling task. To properly reflect the comparable results between each task, I highly recommend that the authors conduct an analysis on the spelling task in the dimension of response times. For instance, as you did for the tracing task, you may calculate the response times to accurately spell the graphemes within the given word. Then, there will be a thorough understanding of the effect of the consistent, inconsistent, and specific to English letter-sound pairings on the spelling task.

5. In the section of Discussion on page 16 lines 573-575, it is stated that “Given these challenges, we expected that additional training methods using a multimodal presentation to support the learning of GPC (grapheme-to-phoneme correspondence) would be beneficial”. This sentence may lead readers to the thought flow that it is difficult and challenging to acquire some types of letter-sound correspondence (e.g. inconsistent and specific-to-English graphemes), and the interactive digital teaching tool may bring about positive outcomes in learning these challenging mappings. However, the follow-up content was centered around interpreting the effect of the app-based reading training on overall performance across all types of letter-sound correspondence, instead of respectively classifying its effect based on different types of English graphemes. 

Therefore, my concern is that the authors can re-write this sentence to make a more proper transition statement to pave the way for the follow-up discussion, and probably also mention the lack of such specific analysis which may be the focus of the further studies.

 

Author Response

please see attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors may find my comments and suggestions for improvement in the attached .pdf file. 

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I found the present version much easier to follow. The data analyses and figures are now clear (even for less/non expert readership). The findings and implications of the current study are thoroughly discussed. I think that the present manuscript represents a nice addition to the literature.

 

Author Response

Thank you for your input and helping us improve the manuscript