Language Dominance Modulates the Perception of Spanish Approximants in Late Bilinguals

Round 1

Reviewer 1 Report

The article investigates the role of dominance in the perception of Spanish approximants by English-Spanish and Spanish-English bilinguals. The use of a composite score of dominance instead of relying on discrete variables (e.g., proficiency) seems timely, as the interest on the effect of dominance on different areas of language learning and use is growing. Since L2 sound perception is among the areas less investigated, extending the focus of study to the perception of L2 approximants offers a valuable contribution to its field.

The main strength of this manuscript is that it provides data for an understudied issue, as hinted in the previous paragraph. In addition, the authors seem to have paid good attention to the research design. Nonetheless, there are a couple of details that require attention. Ideally, (at least) the Spanish-speaking group should be larger. Ten participants seem like a low number to carry statistical analyses. Moreover, some of them have been immersed in an English-speaking environment for a considerable amount of time (see comment 14), which may affect the way they perceive certain sounds. At least in terms of LOR, the Spanish-speaking group seems more heterogeneous than it would be desirable. See below for some additional comments and suggestions for the authors.

Specific comments

Lines 41 to 47

Comment: Too long of a sentence that could be easily divided into two.

Section 1 and its sub-sections (1.2 to 1.6)

Comment: Point 1.1 seems to be missing. See how section 2 (Materials) is organized in a different way (main section heading, then 2.1, 2.2, and so on). I would suggest separating the Introduction (section 1) from the review of Previous Studies (section 2). Alternatively, I would renumber section 1 so that it includes point 1.1 and is more similar to the current section 2.

 

Line 117: , as mentioned above in most varieties of Spanish

Comment: Comma missing after above.

 

Line 125: Zampini, 1994))

Comment: Delete one bracket.

 

Line 148: some other “old” sounds

         Comment: Please, specify.

 

Line 173: length of residence (LOR)

Comment: This acronym has already been introduced (see line 30). No need to spell it out in full here.

 

Line 177: Fox, Flege and Munro

Comment: The authors are alternating “and” and “&” (see, for example, line 75: Face & Menke). For the sake of consistency, they should choose one form and use it throughout the text.

 

Line 187: in L2 perception and the BLP

Comment: May want to spell BLP out in the heading.

 

Line 197: language history, language use, proficiency and attitudes

Comment: The authors may want to add a comma after proficiency.

 

Line 222:

Comment: Since the literature review touches upon several sub-topics, it would be nice for the reader to find a paragraph summarizing and tying up the main ideas from 1.3, 1.4, and 1.5. This would also be a way to further contextualize and motivate the research questions.

 

Hypothesis 2:

Comment: Are both contrasts equally difficult or one of them could be more difficult than the other? Besides basing their hypothesis on Kissling’s findings, the authors could elaborate a bit on the issue.

 

Lines 249-250:

Comment: Given that French is an official language in Canada, could this language be the true L2 (in order of acquisition and/or, more importantly, proficiency) instead of a third language for the 19 participants who report knowledge of French?

Also, do French or Portuguese align with Spanish in terms of the contrasts studied here? This could be a problem if the participants were proficient in French or Portuguese, as it could facilitate perception of such contrasts. Otherwise, it should not be a big deal, in my opinion. For this very same reason (i.e., whether the contrasts exist in other known languages or not), the authors should report all languages spoken by the English-speaking participants besides French and Portuguese.

 

Lines 262-267:

Comment: This title seems too long for a table. Some of the information seems to deserve to be explained in a paragraph within the body of this ms. Probably, details about the BLP can be added to the description of the BLP or to the calculation of scores in the results. In addition, the reader should get details about the near-balanced group somewhere else in the article. As per the numbers reported, we can see that most near-balanced bilinguals come from the L1 Spanish group, but it should not be up to the reader to do the math.

 

Lines 254-255:

Comment: What is the mean LOR? And the range of years L1 Spanish-speaking participants have spent immersed in an English-speaking environment? As per the range of ages (18 to 51) reported, we could suspect some of them have been in Canada for a considerable amount of years (which could also explain why many of the near-balanced group consists mainly of Spanish speakers). How can a long LOR affect perception of the target contrasts? Can there be some attrition? Could participants with a long LOR be less sensitive to the contrast? The “control” group is already small enough (only ten participants) to have to deal with potential differences like this across participants. It would be (or would have been) advisable to: 1) choose only Spanish speakers that had arrived recently, 2) recruit more participants, and 3) run the statistical analyses again.

Alternatively, the authors should report the group size and what seem like large differences in terms of LOR and the small group size as limitations of the study.

 

Line 277:

Comment: Why were the items non-words instead of real words or a mixture of real words and non-words? While there may be advantages to this, the reader should know the reasons behind this choice.

 

Figure 3:

Comment: The upper part of the error bar on the left-most graph (/b/ score) is cut. Also, the authors may consider using colors instead of patterns, since patterns make it more difficult to clearly see the lower part of the bars.

 

Lines 426-435:

Comment: Somewhat repetitive. Wording could be improved.

 

Line 426:

Comment: Comma after Kissling (2015)?

 

Line 449:

Comment: the perceptual assimilation model (Best & Tyler, 2007) should probably be the Perceptual Assimilation Model (PAM; Best & Tyler, 2007). It seems a bit awkward to introduce a model in the discussion section. Would it not be relevant to briefly present this model together with the SLM in section 1.2?

 

Lines 461-464:

Comment: The authors claim that “the present study contributes to the concept of a changing individual language profile [...] throughout the lifetime of late bilinguals”. In order to make such a claim, shouldn’t have they used a longitudinal design?

 

Lines 478-479:

Comment: I would tone down the assertion that these approximants are the sounds that “most contribute to the perception of a foreign accent by native speakers of Spanish”. What about aspiration of word-initial voiceless stops, for example?

 

Reference list:

Comment: There are several problems with the reference list, like incomplete references (see Hualde et al), inconsistent use of & and and (Warner and Tucker vs. Amengual & Simonet, for example), and others.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The present study is focused on the correlation between the language dominance and the perception of auditory phonetic contrasts [β, δ, ɣ] v.s. [b,d,g] through the observation of English-Spanish bilinguals’ performance in AX discrimination task. The findings indicated that the more dominant in Spanish, the more proficient in discriminating the approximant allophones. Additionally, across all language dominance types, there is a hierarchy for perceptual recognition, and the increasing difficulty order is β < δ < ɣ.

In lines 90- 91 of the section 1.3, “Martínez-Celdrán (2013) reports a 96.3% rate of approximants, 2.1% fricatives [β, ð, ɣ], and only 1.7% stop articulations following a vowel” I understand that authors here intend to support the argument that approximant allophones are more likelihood to be articulated as a variant form of voiced stops in Spanish when a vowel precedes the voiced stops. However, what may cause a confusing point is that the phonetic symbols [β, ð, ɣ] are put with “2.1% fricatives”, while in the beginning of this section these symbols are linked with Spanish approximants. Is there any misuse of phonetic symbols here? Does the “2.1% fricatives” mean another variant of voiced stop? Is it a less weakening phonetic articulation compared to approximants? Since it is the first time to mention fricatives in Spanish, an explicit explanation for this is needed here.

 

In lines 91-92, the discussions are shifted to the English phonetic rules with the sentence “It is worth noting that in English, the fricative [ð] and the stop [d] form a phonemic contrast (e.g. [ðo] “though” vs. [do] “dough”).” Again, when the preceding content explicitly explains the fricative allophone articulation in Spanish, the example of English fricative [ð] and the stop [d] can demonstrate its meaning. I suggest that an explanatory statement to be inserted here to increase the connection with the previous phonetic rules in Spanish, such like to clearly point out that the contrastive distance between [ð] and [d] is phonetically larger in English than in Spanish, since this phonetic pair is viewed as allophony in Spanish but as seperate phonemes in English.

 

From the line 123, the authors listed detailed evidences in literature that proved English L2 learners of Spanish showed low performance in the discrimination between approximants and stops, and the authors also indicated that “…but have difficulty with its implementation due to the absence of an obligatory allophonic rule for lenition in English, …”To be coherent with the previously mentioned SLM, it is suggested that several lines could be added here like----These findings to some extent support Speech Learning Model. That is, English and Spanish share a mutual effect in perceiving the sound of voiced stops. The stops in L2 Spanish with similarly articulated allophone variants acoustically resemble to the stops in L1 English, and the equivalence classification causes the transfer of L1 equivalent ……

 

In line 262 of Table 1 in the section of Materials and Methods, I assume that the global dominance is the subtraction of English overall scores from the Spanish ones, but the authors did not explicitly describe this. It is better to clarify the way that global dominance is calculated here.

 

In line 270, the subsection of Tasks, it is important to provide the information such as what software or techniques were used to present AX task? What does the AX discrimination task usually refer to? It is useful to provide a brief introduction of this task.

 

In lines 322 to 323 of section Results, “the following numerical groupings (3) were established:……”. The groupings were set by the index of BLP score, and the criterion was larger than 100, smaller than -100 or between -100 and 100. Is there any statistical meaning for this figure 100? In other words, why could 100 be used as a standard to classify the dominance group, is it chosen arbitrarily or with any statistic reasonings?

 

In lines 325 to 326 of the section Results, “…perceptual accuracy scores were not statistically different between the near-balanced and Spanish-dominant groups.” It is necessary to point out what statistical analysis was conducted here through what statistic software, ANOVA analysis with SPSS? Besides, the p-value should be supplemented here to support this statement “not statistically different”. Similarly, in lines 384 to 386, the pairwise comparisons conducted on each level of each group should also include the report of p-value.

 

In line 347, the authors stated “an independent-samples t-test”. Even though the authors reduced the groups into two, Spanish-dominant (12 participants) v.s. English-dominant (21 participants), the sample size is not sufficiently large. There may be a possibility of violating assumption of the equality of variances, and what is the result for the Levene’s test? If the violations come into place, the non-parametric tests could be used or Welch-corrected t-test could be reported. In other words, the results of this t-test could be reported in more details.

                                                          

In the section of Discussion, the “L2 Spanish learners” (such as lines 420- 421: “…therefore indicating that L2 Spanish learners are in fact in the process of acquiring this perceptual contrast…”) are used to refer to some of the results in the present study. However, according to the preceding content in the section of Participants and Results, it showed that 23 L2 Spanish learners and 10 L1 Spanish speakers were mixed together and grouped into “Spanish-dominant group (n=5), near-balanced group (n=7), and English-dominant group (n=21)”. That is, the term “L2 Spanish learners” cannot be generalized to all participants, since L2 English learners were also present in the grouping and entangled with the data analysis. Therefore, in the deduction of the discussion, could the authors be more concise in the interpretation of the results? In this case, some previous findings which were specifically focused on L2 Spanish learners are not that supportive to the implication that authors proposed, and maybe some other evidences should be supplemented.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

This is a well written paper on a soundly designed experiment. The only comment I have is that the author(s) reported using non-words in the study, but I detected several actual Spanish words in the list, including 'hago', 'ido', 'haba', and 'hada'. This should be addressed within the paper. 

Author Response

Please see the attachment. 

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors have responded to all the questions and no further comments on it.