Comparing Measuring Methods for Speech Timing During Rapid Automatic Naming Tasks Using Arabic and English Bilingual Adults

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

 

REVIEW OF MANUSCRIPT FOR JOURNAL: LANGUAGES

 

Name of manuscript: Comparing Measuring Methods for Speech Timing during 2 Rapid Automatic Naming Tasks using Arabic and English 3 Bilingual Adults

 

This is an interesting manuscript that explores the effect of using three different methods for making utterance rate measurements for a pair of RAN tasks in the participants’ L1 (Arabic) and L2 (English). Three hypothesis were built: 1) The participants’ utterance rates would be slower in their L2. 2) L2 utterances would be marked by longer and more frequent pauses than the L1 ones. 3)  Because of word structure differences between the two languages, that measuring the utterance timing in articulation rate by counting syllables per second would most accurately depict the naming speed among these speakers of two languages. 20 Arabic and English speaking young adults completed RAN tasks for number and object naming in both languages. Utterance rate, speaking rate, and articulation rate were measured. The participants exhibited similar utterance durations in the two languages and utterance rate in words per second, but faster utterance rates in Arabic when measuring the utterance rate in syllables per second. They exhibited more pauses and longer pauses during the English productions, particularly for the object naming. Finally, the speaking rate in syllables per second and articulation rate in syllables per second were equivalent as utterance rates.

Sampling Issues

As described on page 4, they utilized a group of 20 (17 male and 3 female) 19-35 year-old young adult speakers of two languages, Arabic (L1) and English.

There is no real describing of the sampling. No information is provided about SES etc. Why small sample? Why the predominance of males?. Also, there is no clear indication that the sample includes a balanced bilingual group. The authors did not perform reading tests in the two languages to provide clearer justification for testing this group. Additionally, the authors need to provide clearer evidence of assessment of level of bilingualism with regard to language functioning, especially in what pertains to the variables being tested in the study.  Authors also need to clarify the exclusion criteria that seems to be absent. Was there any attempt to assess these children in any way with regard to possible reading disability?

The authors provide a discussion in terms of the inherently different demands in the two languages selected for the study (i.e., Arabic and English) and a good review of the literature to support their research. The manuscript, however, suffers from a lack of clarify on a number of crucial areas that could affect the interpretability of the data. The authors should address the issues listed below so as to make the manuscript clearer before is to be considered for publication. At times, the authors make assumptions and leave things unexplained. For example, they are agnostic about previous publication on RAN in Arabic (see Ibrahim, R. (2015). How does Rapid Automatized Naming (RAN) correlate with measures of reading fluency in Arabic. Psychology. 6 (1), 269-277) adding to the general lack of clarity. In such publication, it was proposed that there is consistent results regarding the relationship between naming speed (NS) and reading ability.

 

Another problem that could limit the conclusion that can be drawn from this study- and the authors mentioned it in the limitation of the study- is about the way in which responses were timed related to L1-L2 differences in the digit naming speeds of bilingual speakers (eg. the use of  talking rate measures for RAN tasks).

 

Author Response

1. Sampling method is better described as a convenience sample.
2. The predominance of males may have been a consequence of the population of Middle Eastern students at Florida State University. For example, during the year of the data collection the Center for Intensive English Studies at Florida State University 15 Arab male students and 4 female Arab students.
3. We did not gather SES information on the participants. However, they were all Middle Eastern natives attending university in the USA. They all appeared to be at least middle class. If they were state supported, they had adequate means for housing, transportation, and all university expenses.
4. We are not certain what the reviewer meant. However, since the participants were all university students who were Middle Eastern, they were all Arabic L1 and English L2 speakers. They represented the population of interest for the study.
5. We were assessing the use of different methods for measuring naming speed. That is why we did not assess their reading abilities.
6. We more thoroughly described our method of determining the participants’ language skills in their second language, English.
7. Every person who responded to our request for participants met the criteria of being a university student who was a native speaker of Arabic from a Middle Eastern country.
8. The participants were all adults who were succeeding as university students in their second language. I believe we can assume that they did not have discernable reading disabilities.
9. We included more information about previous work on using RAN tasks with Arabic speaking populations, including the one by Ibrahim.

10 Rather than investigating the relationship between naming and reading, the purpose of our study was an analysis of different methods to measure speech timing. As a result, we did not address reading fluency. We are aware that RAN tasks are frequently used as a indicator of reading fluency. We hope that those who use RAN tasks for that purpose will heed our results and not use utterance duration as the measure of RAN, but use speaking rate as it accounts for the number of syllables that participants use to complete the RAN tasks.

11. The reviewer’s final issue with the manuscript addresses the purpose of our study, but this statement seems to have a reading focus. Our focus was on how RAN tasks are measured, not on their association with reading fluency.

Reviewer 2 Report

Comments and Suggestions for Authors

This study examined talking speed differences in the utterances of a group of Arabic-English bilinguals. Results showed similar utterance durations in both language and utterance rate in words per second, but faster utterance rates in Arabic when measuring the utterance rate in syllables per second. Overall, the manuscript is clearly written and presents interesting results. However, there are major concerns regarding the theoretical background and methodology that require further justifications. Please see detailed comments below:

 

Major

Introduction:

1.        The authors mentioned that the speaking rates are different between L1 and L2. Does this apply to all types of languages? In which tasks, and from which linguistic perspective? More evidence is needed from previous literature.

2.        The rationale of why speaking rate and articulation rate for measuring RAN may differ between languages remain unclear. Specifically, the authors assume that “the difference is that the method of measuring utterance rates to determine RAN speed may have interacted with the orthographic opacity and reduced the differences across language” (line 139-140) needs further justifications and citations. How do utterance rates interact with orthographic opacity? In which way/direction?

3.        Similar to the above comment, I think the authors should elaborate more in the background about 1) the differences between Arabic and English, 2) why choosing these two languages, 3) how the cross-linguistic differences may influence those rates in measuring RAN between languages? All these points need to be clearly articulated in the background, and should directly connect to each specific hypothesis for the current study (line 166-170).

 

Methods:

1.        Details about the participants are needed. Since they are all university students, how long have they been in the U.S.? The authors reported “for at least eight months”, but what’s the range? This information will highly affect their English proficiency. Also, are any of them immigrants? When did they acquire English, all at early or late age?

2.        More information about the language proficiency measure is needed. What are those CEFR ratings (line 188). Which language skills were these measuring? Speaking, writing, listening, or reading?

3.        How did you classify “independent and proficient users of English” (line 189)? More information needs to be provided about the classification of bilingual participants.

4.        Procedures: in what kind of setting were the participants tested?

5.        Measures: were the Arabic and English tasks counterbalanced across participants? Also, did all participants completed the digit naming task first, followed by the object naming task? If so, the authors should take into the order effect in their data analysis.

6.        Data analysis: Calculating a sample size for data analysis seems like an unusual step to me. Given that your calculated sample size is 18, which is very close to all the 20 participants you collected, why not just including all data in the analysis then?

7.        As mentioned earlier, given that your tasks were not counterbalanced across participants, any factors related to the order, such as practice effect, may significantly influence task performance on the second task, i.e., English in this case. So, in the data analysis, the task order should be included as a covariate. Alternatively, results from the first task should be compared with results from the second to check for any order effect.

Discussion:

1.        The authors need to flesh out more about each finding. For example, your first finding of similar utterance duration in both languages did not support your hypothesis. Why so? Any theoretical evidence may explain this finding?

2.        This study used several different measures. What’s the key message? I think the implications should be further expanded by linking all those results together. Are these three measures of rate comparable, and how important it is to include all of them in future bilingual research?

3.        The authors should discuss how the findings are affected by potential order effect. At least this should be mentioned in the limitations.

4.        How likely are the results generalizable to other bilingual language groups? How will the findings be affected by linguistic differences?

 

Minor:

 

1.        Line 114: should be “different aspects…”?

2.        Figure 2: the color for each condition is unclear, especially that the two EN tasks show similar shading.

Comments on the Quality of English Language

There are several places with odd grammar, please see details in my comments.

Author Response

Introduction

1. The section on speaking rate differences between L1 and L2 has been expanded. We were not certain what the reviewer meant by “from which linguistic perspective”. Hopefully, the expanded review of articles on the speaking rate differences will adequately address the reviewer’s concern.
2. The reviewer states that we assumed “the difference is that the method of measuring utterance rates to determine RAN speed may have interacted with the orthographic opacity and reduced the differences across languages”. However, the sentence began “A possible reason for these differences”. We stated this as a speculation, not as an assumption. The section has been rewritten to clarify what we say and we removed the clause that was indicated.
3. We expanded the section on Arabic in the introduction. We also expanded on some differences between Arabic and English that are relevant for this study. We also made clearer the reason for contrasting English, the language used in most RAN studies, and Arabic as a foundation for the need to use other methods to determine speaking rates for RAN tasks.

 

Methods

1. The description of the participants is more thorough. None of them were immigrants, as all intended to return to their home nations after they completed their education. Their early learning of English was in school as added to the manuscript.
2. The CEFR ratings are more thoroughly described and a table is included to help clarify these ratings.
3. The reviewer’s question was addressed in our response to Methods item #2. The words quoted were from the CEFR ratings. The table and text expand on the words to help make the concept clear.
4. The room used for the research tasks is described.
5. As noted in the manuscript, the tasks were done in Arabic first in order to increase the participants’ comfort in completing them. As the speaking and articulation rates in English were slower than those in Arabic, it appears that they did not do better on the English portion of the tasks as would be expected if an order effect had occurred.
6. We calculated the sample size to determine how many participants we needed in order to use parametric statistics. That way we knew that we had enough participants when we reached 20.
7. As noted in our response to Methods item #5, there was no order effect. Therefore, there was no need to consider it as a covariate.

 

Discussion

 

1. The discussions in all sections are expanded.
2. The key message that speaking rate is highly effective for measuring RAN tasks across languages is more clearly stated in sections 4.4 and 4.5.
3. Order effect is mentioned in the Limitations of the study.
4. We expanded our discussion to report that Georgiou et al (2022) also reported using speaking rate an accurate method for measuring RAN tasks. They had speakers of six languages from three language families. Since it more accurately reflects the actual utterances used by participants, it should be the most accurate methods regardless of language.

 

 

 

Minor

 

1. Fixed
2. The fill in the boxes for the digit naming durations in Figure 2 were redone to more clearly differentiate them across languages and to differentiate them from the object naming durations.