Datasets:
File size: 124,608 Bytes
0f2f2d3 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 | {
"v1_Abstract": "Working memory (WM) is often poorer for a second language (L2). In low noise conditions, people listening to a language other than their first language (L1) may have similar auditory perception skills for that L2 as native listeners, but do worse in high noise conditions and this has been attributed to the poorer WM for L2. Given that WM is critical for academic success in children and young adults, these speech in noise effects have implications for academic performance where the language of instruction is L2 for a student. We used a well-established Speech-in-Noise task as a verbal WM (vWM) test and developed a model correlating vWM and measures of English proficiency and/or usage to scholastic outcomes in a multi-faceted assessment medical education program. Significant differences in Speech-Noise Ratio (SNR50) values were observed between medical undergraduates who had learned English before or after five years of age, with the latter group doing worse in the ability to extract whole connected speech in the presence of background multi-talker babble (Student-t tests, p <0.001). Significant negative correlations were observed between the SNR50 and seven of the nine variables of English usage, learning styles, stress, and musical abilities in a questionnaire administered to the students previously. The remaining two variables, Perceived Stress Scale (PSS) and the Age of Acquisition of English (AoAoE) were significantly positively correlated with the SNR50, showing that those with a poorer capacity to discriminate simple English sentences from noise had learnt English later in life and had higher levels of stress \u2013 all characteristics of the international students. Local students exhibited significantly lower SNR50 scores and were significantly younger when they first learnt English. No significant correlation was detected between the SNR50 and the students\u2019 Visual/Verbal Learning Style (r=-0.023) . Standard multiple regression was carried out to assess the relationship between language proficiency and verbal working memory (SNR50) using 5 variables of L2 proficiency with the results showing that the variance in SNR50 was significantly predicted by this model (r2=0.335). Hierarchical multiple regression was then used to test the ability of three independent variable measures (SNR50, age of acquisition of English and English proficiency) to predict academic performance as the dependent variable in a factor analysis model which predicted Pre Prin ts Pre Prin ts significant performance differences in an assessment requiring communications skills (p =0.008), but not on a companion assessment requiring knowledge of procedural skills , or other assessments requiring factual knowledge. Thus, impaired vWM for an L2 appears to affect specific communications-based assessments in university medical students.[b] Pre Prin ts Pre Prin ts",
"v1_col_introduction": "introduction : In medical education, most information is communicated verbally, often\nto large groups of students. Consequently, listening abilities and language comprehension are critical to learning and require both auditory perception and auditory working memory (WM) skills. WM is defined as \u201cthe system for the temporary maintenance and manipulation of information, necessary for the performance of such complex cognitive activities as comprehension, learning, and reasoning...\u201d (Baddeley, 1992, p. 281). One core element of WM, and in particular verbal Working Memory (vWM), is the \u201cphonological loop\u201d, which has been shown to be critical for language acquisition during development, as well as language processing in daily life (Baddeley, 1992). However, it has been widely reported that WM capacity may be limited for students who are learning in an environment where the language of instruction is not their native language (Andersson, 2010; Kroll, et al. 2002; Mackey, et al. 2002; McDonald, 2006; Miyake & Friedman, 1998; Service, 1992; Service, et al. 2002; Sunderman & Kroll, 2009; Tokowicz, Michael, & Kroll, 2004) and this appears to be due to demands on WM resources in the second language (L2) (Service, et al., 2002).\nThe relationship between WM capacity and academic achievement has\nbeen well studied in children (Alloway & Elsworth, 2012; Gathercole & Pickering, 2000a, 2000b; Gathercole, et al. 2004; Vock & Holling, 2008) and in university students and adults (Daneman & Carpenter, 1980; Daneman & Hannon, 2001; Swanson, 1994; Tolar, Lederberg & Fletcher, 2009). Whilst the studies in younger learners have shown strong correlations between WM and high academic attainment (Alloway & Alloway, 2010; Gathercole, et al. 2004; St Clair-Thompson & Gathercole, 2006), studies of university science students have reported that WM has only weak or indirect effects in predicting academic performance (Krumm, Ziegler & Buehner, 2008; Rohde & Thompson, 2007). Tolar, et al. (2009) found WM strongly related to the adults\u2019 ability on Scholastic Aptitude Test (SAT) scores, but effects were reduced when other cognitive factors were controlled for, such as spatial ability. Further, some studies suggest that vWM may not have as great an effect on the students\u2019 processing abilities as the direct effects of the students\u2019 first language (L1), including the ability to\n33\nPre Prin ts Pre Prin ts\nsuppress L1 influences or the level of L1 proficiency and general language aptitude (for review see Juffs & Harrington, 2011).\nIn addition or in consequence of the poorer vWM for L2, the acoustic\nenvironment to facilitate ideal listening conditions may also be crucial for effective learning by L2 medical undergraduates. It has been noted that non-native listeners may have similar speech perception skills as native listeners in low noise conditions, but that these abilities significantly decrease in high noise conditions (Buus, et al. 1986; Florentine, et al. 1984; Lin, Chang & Cheung, 2004; Mayo, Florentine & Buus, 1997; Tabri, Abou Chacra & Pring, 2011; Takata & Nabelek, 1990). Using the Speech-in-Noise (SiN) task, Mayo, et al. (1997) showed that not only was speech perception in noise poorer in L2 learners, but that it was also dependent on the age the L2 was acquired; bilinguals who learnt English after 14 years of age had the worst performance in the SiN task compared to monolinguals and bilinguals who learnt English before 6 years of age. Further, in contrast to the monolinguals, the late bilinguals did not benefit from contextual cues in those sentences that were highly predictive (i.e. sentences in which the subjects could easily guess the target word). Similarly, Buus, et al. (1986) found that the noise tolerance level of non-native listeners to understand 50% of the test sentences, increased with years of exposure to English, but never reached the level of tolerance (and achievement) of a native English speaker.\nThere is evidence that the ability to process speech in noise influences\nthe ability to recall academic material. Ljung, et al. (2010) tested 48 native Swedish university students with open-ended questions about the content of spoken lectures of up to eight minutes presented in broadband noise or quiet, or presented students with 10 paragraphs of lectures in classrooms of differing reverberation times. The subjects\u2019 memory performance was significantly worse under both adverse conditions compared with the quiet condition, even when the students had heard correctly the spoken lectures.\nGiven the relationship between vWM capacity, academic achievement\nand the impairment of speech comprehension in noisy environments by L2 learners, such effects are likely to be even stronger for these students. Thus, a\n44\nPre Prin ts Pre Prin ts\npotential disadvantage exists for medical students learning a course in their L2. This is particularly relevant to the many international medical students that travel to mainly English-speaking western universities in, e.g., Australia, the UK or the USA (Brisset, et al. 2010) especially those for whom the L2 was not acquired at an early age. Our study has important implications in identifying another significant factor impacting on the academic performance in the early years of a medical undergraduate course, the period of greatest stress and of greatest likelihood of drop-outs/failures (Baker, 2004).\nIn the present study we examined the relationships between vWM for L2,\nthe age at which the L2 was acquired, and students\u2019 scholastic outcomes. In a previous study (Mann, et al. 2010), we showed that international students in a Bachelor of Medicine/Bachelor of Surgery (MBBS) course in an Australian university performed worse than their local peers, but that this was significantly influenced by the students\u2019 L1. This is consistent with the idea that L1 influences may affect academic outcomes for instruction in an L2. Building on this, we now explore whether verbal WM plays a role in the academic achievements of a cohort of international and local medical undergraduates in the same course. Specifically, we hypothesise that 1) students with English as a Second Language (ESL students) will have lower scores than students with English as a First Language (EFL students) in the SiN test (reflecting poorer vWM); and 2) that the students with lower SiN results will also have lower academic scores in their different assessments.\nAs well as having a high secondary school result (a pre-requisite also for\nlocal students), international medical students must pass stringent measures of English proficiency prior to enrolment and must also attend and pass an interview to demonstrate high motivation and self-expectations. To a major extent these requirements obviate the confounding effects of English proficiency skills often suggested (Lun, Fischer & Ward, 2010; Webb, 2002) to account for the fact that, generally, international medical students do not perform as well academically as their local counterparts (Bagot et al., 2005; Liddell & Koritsas, 2004; Wass, et al. 2003). We used a well-established auditory test paradigm as a vWM test, free of L2 proficiency concerns that have been raised against such\n55\nPre Prin ts Pre Prin ts\ntests as the Reading Span Test (RST) when applied to L2 learners (Juffs & Harrington, 2011). The SiN task tests vWM via the phonological loop through storing, processing and recall of speech in background noise.\n66\nPre Prin ts Pre Prin ts",
"v1_col_limitations": "limitations of the study : We have discussed our findings in relation to verbal WM as the SiN task\nis a verbal/auditory task and, therefore, a measure of the phonological loop of WM. We did not employ visual memory tasks, e.g. written examinations, and further research into how the mode of presentation could affect outcomes is required.\nFurther, we had categorised our AoAoE group as having acquired\nEnglish either before or after the age of 5 years old according to extant literature. In our sample, the age range was 1-12 years, meaning that the majority of subjects in our sample learnt English pre-puberty. Most studies find greater discrepancies with L2 learners who have learnt English post-puberty (~14 years old e.g. Mayo, et al., 1997). Therefore, our results may underestimate the true effect of L2 age of acquisition on advanced learning.\nConclusions and implications for future pedagogical design of MBBS courses\nIn summary, our study contributes to the growing research examining\nwhy non-native medical undergraduates generally perform academically worse than their native speaker counterparts despite having good L2 proficiency skills. The implications are that in a prestigious course such as the MBBS degree, where all students have proven high academic abilities, motivation and expectations prior to commencement, small differences at the early stages could have disproportionate impacts on the medical careers of L2 students, for example, in selection for highly competitive specialist training positions or fellowships. The knowledge from this study, therefore, could be used in the\n2323\nPre Prin ts Pre Prin ts\ntraining of medical students from diverse backgrounds, for instance, by introducing compulsory language immersion programs prior to commencement of the formal course. An immersion program is typically 3-6 months and forces the student to speak and think in the host country\u2019s language in order to understand the language and the culture. Even for students who have apparently high levels of English proficiency (as gauged for our medical students by the stringent IELTS / TOEFL tests and face-to-face interviews) such immersion programs may prove to improve vWM in the language of instruction simply through more extensive use. This could be either general language immersion, or may be better if targeted to the specific clinical and health sciences language that medical students will encounter on commencement of the course. Further, advanced technology could be installed in areas of high noise conditions, e.g. audio systems in lecture theatres, that filter out \u2018white noise\u2019 to give better signal enhancement and brain processing of information to students. Having this information could also help medical students\u2019 in forming appropriate study habits such as understanding what is a \u2018good\u2019 study environment, etc.\nWe note that our study highlights an area where international medical\nstudents continually fall down despite rigorous processes and comparable English proficiency. Under these circumstances, we believe that our study provides a strong basis for carrying out procedures as noted above to improve equity of access by international students to resources to improve their academic outcomes.\n2424\nPre Prin ts Pre Prin ts",
"v2_Abstract": "Working memory (WM) is often poorer for a second language (L2). In low noise conditions, people listening to a language other than their first language (L1) may have similar auditory perception skills for that L2 as native listeners, but do worse in high noise conditions and this has been attributed to the poorer WM for L2. Given that WM is critical for academic success in children and young adults, these speech in noise effects have implications for academic performance where the language of instruction is L2 for a student. We used a well-established Speech-in-Noise task as a verbal WM (vWM) test and developed a model correlating vWM and measures of English proficiency and/or usage to scholastic outcomes in a multi-faceted assessment program. Our model predicted significant performance differences in an assessment requiring communications skills, but not on a companion assessment requiring knowledge of procedural skills or other assessments requiring factual knowledge. Thus, impaired vWM for an L2 appears to affect specific communications-based assessments in tertiary medical students.",
"v2_col_introduction": "introduction : In medical education, most information is communicated verbally, often\nto large groups of students. Consequently, listening abilities and language comprehension are critical to learning and require both auditory perception and auditory working memory (WM) skills. WM is defined as \u201cthe system for the temporary maintenance and manipulation of information, necessary for the performance of such complex cognitive activities as comprehension, learning, and reasoning...\u201d (Baddeley, 1992, p. 281). One core element of WM, and in particular verbal Working Memory (vWM), is the \u201cphonological loop\u201d, which has been shown to be critical for language acquisition during development, as well as language processing in daily life (Baddeley, 1992). However, it has been widely reported that WM capacity may be limited for students who are learning in an environment where the language of instruction is not their native language (Andersson, 2010; Kroll, Michael, Tokowicz, & Dufour, 2002; Mackey, Philp, Egi, Fujii, & Tatsumi, 2002; McDonald, 2006; Miyake & Friedman, 1998; Service, 1992; Service, Simola, Mets\u00e4nheimo, & Maury, 2002; Sunderman & Kroll, 2009; Tokowicz, Michael, & Kroll, 2004) and this appears to be due to demands on WM resources in the second language (L2) (Service, et al., 2002).\nThe relationship between WM capacity and academic achievement has\nbeen well studied in children (Alloway & Elsworth, 2012; Gathercole & Pickering, 2000a, 2000b; Gathercole, Pickering, Ambridge & Wearing, 2004; Vock & Holling, 2008) and in university students and adults (Daneman & Carpenter, 1980; Daneman & Hannon, 2001; Swanson, 1994; Tolar, Lederberg & Fletcher, 2009). Whilst the studies in younger learners have shown strong correlations between WM and high academic attainment (Alloway & Alloway, 2010; Gathercole, Pickering, Knight & Stegmann, 2004; St Clair-Thompson & Gathercole, 2006), studies of university science students have reported that WM has only weak or indirect effects in predicting academic performance (Krumm, Ziegler & Buehner, 2008; Rohde & Thompson, 2007). Tolar et al (2009) found WM strongly related to the adults\u2019 ability on SAT scores, but effects were reduced when other cognitive factors were controlled for, such as spatial ability. Further, some studies suggest that vWM may not have as great an effect on the students\u2019 processing abilities as the direct effects of the\n44\n5\n62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79\n80 81 82 83 84 85 86 87 88 89 90 91 92 93 94\n6\nPre Prin ts Pre Prin ts\nstudents\u2019 first language (L1), including the ability to suppress L1 influences or the level of L1 proficiency and general language aptitude (for review see Juffs & Harrington, 2011).\nIn addition or in consequence of the poorer vWM for L2, the acoustic\nenvironment to facilitate ideal listening conditions may also be crucial for effective learning by L2 medical undergraduates. It has been noted that non-native listeners may have similar speech perception skills as native listeners in low noise conditions, but that these abilities significantly decrease in high noise conditions (Buus, Florentine, Scharf & Canevet 1986; Florentine, Buus, Scharf, & Canevet 1984; Lin, Chang & Cheung, 2004; Mayo, Florentine & Buus, 1997; Tabri, Abou Chacra & Pring, 2011; Takata & Nabelek, 1990). Using the Speech-in-Noise (SiN) task, Mayo et al. (1997) showed that not only was speech perception in noise poorer in L2 learners, but that it was also dependent on the age the L2 was acquired; bilinguals who learnt English after 14 years of age had the worst performance in the SiN task compared to monolinguals and bilinguals who learnt English before 6 years of age. Further, in contrast to the monolinguals, the late bilinguals did not benefit from contextual cues in those sentences that were highly predictive (i.e. sentences in which the subjects could easily guess the target word). Similarly, Buus et al. (1986) found that the noise tolerance level of non-native listeners to understand 50% of the test sentences, increased with years of exposure to English, but never reached the level of tolerance (and achievement) of a native English speaker.\nThere is evidence that the ability to process speech in noise influences\nthe ability to recall academic material. Ljung et al (2010) tested 48 native Swedish university students with open-ended questions about the content of spoken lectures of up to eight minutes presented in broadband noise or quiet, or presented students with 10 paragraphs of lectures in classrooms of differing reverberation times. The subjects\u2019 memory performance was significantly worse under both adverse conditions compared with the quiet condition, even when the students had heard correctly the spoken lectures.\nGiven the relationship between vWM capacity, academic achievement\nand the impairment of speech comprehension in noisy environments by L2\n55\n7\n95 96 97\n98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116\n117 118 119 120 121 122 123 124\n125 126\n8\nPre Prin ts Pre Prin ts\nlearners, such effects are likely to be even stronger for these students. Thus, a potential disadvantage exists for medical students learning a course in their L2. This is particularly relevant to the many international medical students that travel to mainly English-speaking western universities in, e.g., Australia, the UK or the USA (Brisset, Safdar, Lewis & Sabatier, 2010) especially those for whom the L2 was not acquired at an early age.\nIn the present study we examined the relationships between vWM for L2,\nthe age at which the L2 was acquired, and students\u2019 scholastic outcomes. In a previous study (Mann, Canny, Lindley & Rajan, 2010), we showed that international students in a Bachelor of Medicine/Bachelor of Surgery (MBBS) course in an Australian university performed worse than their local peers, but that this was significantly influenced by the students\u2019 L1. This is consistent with the idea that L1 influences may affect academic outcomes for instruction in an L2. Building on this, we now explore whether verbal WM plays a role in the academic achievements of a cohort of international and local medical undergraduates in the same course. International medical students must pass stringent measures of English proficiency prior to enrolment and must also attend and pass an interview to demonstrate high motivation and self-expectations. To a major extent these requirements obviate the confounding effects of English proficiency skills often suggested (Lun, Fischer & Ward, 2010; Webb, 2002) to account for the fact that, generally, international medical students do not perform as well academically as their local counterparts (Bagot et al., 2005; Liddell & Koritsas, 2004; Wass, Roberts, Hoogenboom, Jones & Vleuten, 2003). We used a well-established auditory test paradigm as a vWM test, free of L2 proficiency concerns that have been raised against such tests as the Reading Span Test (RST) when applied to L2 learners (Juffs & Harrington, 2011). The \u2018Speech-in-Noise\u2019 or \u2018SiN\u2019 task tests vWM via the phonological loop through storing, processing and recall of speech in background noise.\n66\n9\n127 128 129 130 131 132\n133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155\n156\n10\nPre Prin ts Pre Prin ts",
"v2_col_limitations": "limitations of the study : We have discussed our findings in relation to verbal WM as the SiN task\nis a verbal/auditory task and, therefore, a measure of the phonological loop of WM. We did not employ visual memory tasks, e.g. written examinations, and further research into how the mode of presentation could affect outcomes is required.\nFurther, we had categorised our AoAoE group as having acquired\nEnglish either before or after the age of 5 years old according to extant literature. In our sample, the age range was 1-12 years, meaning that the majority of subjects in our sample learnt English pre-puberty. Most studies find greater discrepancies with L2 learners who have learnt English post-puberty (~14 years old e.g. Mayo, et al., 1997). Therefore, our results may underestimate the true effect of L2 age of acquisition on advanced learning.",
"v1_text": "results : Speech in Noise performance and relationship to English proficiency We used the SiN task to assess the presence of vWM deficits in L2 in our medical student population. In comparing across groups, students who had learnt English as a first language, had significantly smaller SNR50 values than the students who had learnt English as a second language (Student\u2019s-t(76) = -4.208, p<0.001) as seen in Figure 1. Twenty-five students were not included in this analysis, as they had learnt English and another language concurrently (true bilingual) and thus did not have English as a first or second language. **Insert Figure 1 here These observations established that the point of subjective performance (the SNR50) from our SiN task is a good index of verbal working memory for L2 in our medical student population. We then used correlational analysis to assess the relationship between SNR50 and English usage items from the questionnaire, as outlined in Table 2. **Insert Table 2 here Significant negative correlations were observed between seven of the nine variables on the questionnaire and the SNR50. The remaining two variables, Perceived Stress Scale (PSS) and the Age of Acquisition of English (AoAoE), were significantly positively correlated with the SNR50, indicating that those with a higher SNR50 ratio (poorer capacity to discriminate simple English sentences from noise) had learnt English later in life, i.e. more likely the international medical students, and had higher levels of stress (as noted in the current literature). Local students exhibited significantly lower SNR50 scores than the international medical undergraduates (t(101) = 6.23, p<0.001), as well as being significantly younger when they first learnt English (t(101) = 3.33, p=0.001). No significant correlation was detected between the SNR50 and the students\u2019 Visual/Verbal Learning Style (r=-0.023), suggesting that the possible cultural variability in this factor was not a substantial confound in our findings. 1414 Pre Prin ts Pre Prin ts On the basis of these observations, we then conducted multiple regression analyses using the five items significantly correlated to SNR50 that pertained to English proficiency and/or usage. These variables were: Age of Acquisition of English (AoAoE); Perceived English Proficiency (PEP); how often their mother (primary caregiver) spoke English when the student was growing up (MSE); the students\u2019 own preference for speaking English (PSE); and how often the student spoke English in the last month (ESLM). All variables were entered simultaneously using the Enter method. The results showed that the variance in SNR50 was significantly predicted by this model of L2 proficiency (F(5, 93) = 9.37, p<0.001, r2=0.335), with the five variables altogether explaining 33.5% of the total variance in SNR50. There were two variables that significantly contributed to this overall variance. The first, Perceived English Proficiency (PEP), had the highest beta coefficient of -0.409 (p<0.001) and accounted for 9.8% of the variance. The other variable was MSE with a beta coefficient of -0.366 (p=0.005) and a unique contribution of 5.91% to the overall 33.5% variance. The other three variables, AoAoE, PSE and ESLM, were not significant predictors of SNR50 in this particular model with beta values of 0.020, 0.159 and -0.019 respectively. However, AoAoE and ESLM showed significant correlations with SNR50. Figure 2 graphically shows the zero-order correlations and beta coefficients for the four variables that were highly correlated to SNR50 as also shown in Table 2. **Insert Figure 2 here One caveat to interpretation of our results is that the five variables pertaining to English proficiency and usage (AoAoE, PEP, MSE, PSE and ESLM) are also highly significantly correlated with each other, with r values >0.5 (Table 2). This may suggest that these variables share the same set of underlying causal elements that affect vWM for L2 and its usage, i.e. they demonstrate multicollinearity. Therefore, a principal component analysis was performed to establish if there were underlying common constructs involved across these factors. The analysis yielded one factor with an eigenvalue >1.0 that accounted for 65% of the variance. All variables had high loadings with a 1515 Pre Prin ts Pre Prin ts minimum of 0.725, and a reliability test yielded a Cronbach\u2019s \u03b1 coefficient of 0.760 (considered an acceptable value of good internal consistency). In order to include all variables in this construct, it is necessary for all variables to be of the same scale. One variable, AoAoE, however, could not be changed (reverse coded) to the same scale as the other four variables in an appropriate way that did not change its correlation values. Therefore, it could not sit in this new construct and as it has been widely documented that language proficiency is influenced by the age at which the language is acquired, hierarchical analysis was conducted. The new construct of the four remaining variables, i.e. PEP, MSE, PSE and ESLM, was representative of the amount of exposure and usage the students had of English and a self-rating of their English skills. It was thus an approximation of the students\u2019 overall English proficiency, renamed \u2018English Language Skills\u2019 (ELS) and the means were calculated for analysis and checked for multicollinearity against SNR50. Hierarchical multiple regression analyses were used, controlling for AoAoE in the first step and SNR50 and the new construct ELS in the second step. Analysis was performed for the End of Year Total scores, as well as for each Assessment (as described in the Methods section) for Year 1 and Year 2 of study. Results are set out in Table 3 and discussed in detail below. **Insert Table 3 here These results establish that not only is SNR50 a good index of verbal working memory for L2, but it could be employed to test if poorer L2 vWM is a strong predictor of academic performance along with language proficiency skills. Academic performance and relationship to English language skills In the first year of study, the results showed that SNR50 and ELS were not significant predictors of overall academic performance, even when AoAoE was controlled for. However, the L2 vWM index (SNR50) did make a significant unique contribution to the OSCE Communications performance, with a beta coefficient of -0.231 (p=0.043). This demonstrated that the smaller the SNR50 1616 Pre Prin ts Pre Prin ts ratio (i.e., the better the vWM for discrimination of simple English sentences from noise), then the greater the Communications score. In contrast to this, results for the OSCE Technical skills showed significant positive correlations with the AoAoE (beta coefficient of 0.326, p=0.023) and with ELS (beta coefficient of 0.329, p=0.030). These correlations showed that students who had learnt English significantly later in life, but who rated their English skills more highly (international students with good English proficiency skills), performed better in the technical aspects of the OSCEs, despite learning the L2 at a later age. The SNR50 was not significant, indicating that L2 vWM does not influence academic performance for this particular assessment. Overall, after controlling for the age English was acquired, there was no clear, major predictor of academic performance in Year 1. In Year 2, this model of vWM and ELS while controlling for AoAoE was a significant predictor of academic performance of the OSCE Communications skills (p=0.008), explaining 21% of the variance of this assessment. ELS had the highest beta coefficient of 0.315 but this was not statistically significant and accounted for only 3.46% to the overall 21% variance. There was also a significant negative correlation with AoAoE on its own in Step 1 (beta coefficient = -0.384, p=0.004), but AoAoE was no longer uniquely significant in the overall model for predicting OSCE Communication skills, indicating it has only an indirect influence on predicting performance of this academic assessment. With regard to the OSCE Technical assessment for Year 2, the effects were incongruous with those observed in the results obtained for Year 1, with the SNR50 now significantly correlated (beta coefficient = 0.346, p=0.038), but AoAoE and ELS showing no correlation with academic performance. As it was the international medical students who exhibited higher SNR50 ratios, this would indicate that these students could be performing better in this category than their local counterparts. This was confirmed by an independent samples t-test, which showed that the international medical students performed better in this assessment in Year 2 than their local peers (t(43.73) = 3.376, p=0.002). This 1717 Pre Prin ts Pre Prin ts would suggest that the international students\u2019 L2 vWM is not impaired in this assessment in Year 2 (as in Year 1), perhaps because the recall of technical data is not as challenging on vWM capacity as conceptual and abstract comprehension (Van Merri\u00ebnboer & Sweller, 2010). Overall, the model is not a significant predictor for this assessment and explains only 11% of the variance, with SNR50 uniquely contributing 8.07%. Although the model was not a significant predictor of the academic performance of the 2nd year total OSCE (i.e. not subdivided into OSCE Communications and OSCE Technical), it is worth noting that it accounts for 13% of the overall variance for this variable, which in the classroom would be regarded as a considerable proportion. T-test analysis of the Year 2 OSCE scores showed that while there was no significant difference between local and international medical students (p=0.113), there was a significant difference for the AoAoE, with students who acquired English before the age of five having better overall marks for the OSCE assessment than those who acquired English later (t(52) = 2.038, p=0.047). This is also evident in the significant negative correlation of AoAoE in Step 1, with a beta coefficient of -0.320 and significant p-value of 0.018. However, in Step 2, AoAoE was no longer significant, demonstrating that there are overlapping effects with the other variables. To summarise, after controlling for the age at which English was first learnt, verbal working memory for English (as indexed by the SNR50 in our speech-in-noise task) and ELS were not strong predictors of the overall End of Year Totals or for the individual Assessments, with the exception of the OSCEs. For the OSCE assessments, the contribution made to the variance by each predictor varied for the OSCE types and was different for each year of study. The OSCE Communications was the only significant model, which in itself is a significant finding and which is discussed later. 1818 Pre Prin ts Pre Prin ts discussion : The relationship between verbal Working Memory and academic attainment has been well documented in L1, particularly with young learners (Gathercole, et al., 2004). However, the role of vWM in predicting academic achievement in L2 adults, particularly medical students, has been only occasionally examined with inconsistent effects (see Harrington & Sawyer, 1992; Juffs & Harrington, 2011). The aim of the current study was to explore if L2 vWM plays a role in academic attainment in ESL students. We indexed L2 vWM using a SiN task as a WM verbal test, as such tasks have been well documented to be a good indicator of L2 vWM and because such a task reflected, to a consistent degree, the background conditions occurring in some of the venues in which information was imparted to student doctors in their course. Linguistically, English target speech and English speech noise consist of many common properties (e.g., phonemes, syllable structures, prosodic features, etc.), which may make it more difficult for listeners, particularly non-native, to segregate target language from background noise and this may contribute to greater informational masking (e.g., Bronkhorst, 2000; Brungart, 2001; Brungart et al., 2001; Lutfi, 1990; Rheebergen and Versfeld, 2005; Scott et al., 2004; Simpson and Cooke, 2005; Van Engen, 2010; Van Engen and Bradlow, 2007). Background masking noise can be classed as energetic or informational; energetic masking is thought to affect speech processing at the level of the auditory periphery, whereas informational masking, e.g. babble noise, interferes with higher-order processing such as attention and cognitive load. Informational maskers have therefore been often used in working memory tasks to good effect. Hygge, et al. (2003) found that meaningful irrelevant speech noise significantly impaired recall in a text-reading memory task in 92 native high school students in Sweden. We also examined a number of other factors known or postulated to influence L2 skills, in particular the age at which the participants first learnt English (as their L2) as this factor has previously been shown to influence English learning and proficiency (Johnson & Newport, 1989). 1919 Pre Prin ts Pre Prin ts In our first analysis, we confirmed that the point of subjective performance (the SNR50 score) in our SiN task was indeed a good index of verbal working memory for L2 in our student population, with our results showing that the EFL students had smaller SNR50 scores than the ESL students. This meant that the EFL medical students were better able to identify simple English words in a noisy background than the ESL medical students. This was an important step as this SiN task is free of L2 proficiency concerns that have been a major criticism of previous studies that have used measures such as the Reading Span Task (Harrington & Sawyer, 1992; Juffs & Harrington, 2011) to show differences in L2 vWM and may be one explanation for the mixed findings of past studies. It is also worth noting that Waters & Caplan (2005) have argued that traditional measures of WM do not relate to on-line processing of sentences which they postulate to be due to a specialised WM system; we believe that tasks such as the SiN task are likely to be better evaluators of WM in online processing of whole connected speech. We then used this index of vWM along with English Language Skills (ELS) as our model to predict academic attainment whilst controlling for the age that English was first acquired by the student (AoAoE). Different language-related factors affect different subcategories of the Objective statistical analyses : Statistical analyses were performed using SPSS v19.0.0 (SPSS Statistics Inc.) for Windows. All statistical tests were parametric, and data were checked for normality of distribution and variation. Pearson\u2019s correlation was conducted to investigate the relationship between items from the questionnaire, Perceived Stress Scale, Index of Learning Style (the visual/verbal component only was analysed as the other components are not pertinent to this particular study) and Signal to Noise Ratio (SNR50). Standard multiple regression was carried out to assess the relationship between language proficiency and verbal working memory (SNR50) and hierarchical multiple regression was used to test 1212 Pre Prin ts Pre Prin ts the ability of three measures (SNR50, age of acquisition of English and English proficiency) to predict academic performance. Student\u2019s t-tests were also used when comparing independent groups. **Insert Table 1 here. 1313 Pre Prin ts Pre Prin ts material and methods : participants : All participants in this study were students enrolled in the MBBS program from 2008-2010 at Monash University. The students were informed that this project was biphasic and participation involved both completing a questionnaire and an invitation at a later date to undergo an audiometry test. The questionnaire asked for information on the students\u2019 personal demographics, English acquisition and usage, musical abilities and two psychometric measures: Perceived Stress Scale (Cohen, 1994) and the Index of Learning Styles Questionnaire (Felder & Soloman, 1994). Stress has been found to have a negative impact on the academic performance of first year medical students, particularly international students (Bagot, et al., 2005; Baker, 2004; Lacina, 2002; Mori, 2000) as well as the style of learning adopted by international versus local students, such as deep vs. surface learning styles (Bagot et al., 2005; Newble & Entwistle, 1986; Volet, Renshaw, & Tietzel, 1994; Zeegers, 2001). As mentioned in the Introduction, the international medical students of this course must pass stringent measures of English proficiency prior to enrolment, such as the International English Language Testing System (IELTS) or the Test Of English as a Foreign Language (TOEFL). Therefore, the questions on the survey pertained mainly to measurable English attributes such as \u2018In what order did you learn English and your other language\u2019? There was one question on the students\u2019 perceived English and Language Other Than English (LOTE) proficiency which was purely self-rated from a score of \u20180=poor\u2019 to \u20184=excellent\u2019. The surveys were distributed at the commencement of each university year in the 1st year of the medical undergraduates\u2019 course. Of the 791 questionnaires distributed over the three years, 582 were returned giving a response rate of 73.6%. Participation was voluntary and students could withdraw at any stage. In the second phase of the project, students were asked to participate in a SiN test (described below). As it was not feasible to submit all 582 subjects to this test, we performed a power analysis using GPower 3.0.10, which calculated 77 Pre Prin ts Pre Prin ts that we would require 15 subjects in each group to give us an effect size of 0.8 at a power level of 90%. We then emailed all 582 students inviting them to attend the audiometry test at a mutually convenient time. From these emails, we had a total of 113 subjects that came in to be tested on the speech-in-noise task. Of these 113, ten participants were excluded from data analysis: one subject was excluded due to hearing impairments and nine candidates were classed as outliers with means more than two standard deviations from the sample mean (at \u03b1=0.05), leaving a total of 103 subjects tested and analysed, which still gave us ample power for this particular study. Analysis and findings relevant to all 582 students (including the 113 who participated in the audiometry tests) are currently being researched by the authors, and will be reported elsewhere; the emphasis of this report is on the outcomes of the 103 subjects undertaking the SiN test. Demographic characteristics are set out in Table 1. Students were classed as \u2018local\u2019 if they were Australian or New Zealand citizens, or if they held permanent residency for more than three years; or students were classed as \u2018international\u2019 if they held temporary entry visas, in accordance with the option chosen by the students on their questionnaires. Only one student held permanent residency status and had been living in Australia for over five years; all other students were citizens or held temporary entry visas. All ethics for this study were approved by the Monash University Human Research Ethics Committee (MUHREC). audiometry testing : At the outset, hearing sensitivity in each subject was measured with audiometry using a Beltone Model 110 Clinical Audiometer, calibrated to present pure tones through calibrated TDH headphones. Hearing was tested one ear at a time at 500Hz, 1000Hz, 2000Hz, 4000Hz, 6000Hz and 8000Hz. The minimum sound level at each frequency was recorded as the threshold in decibels Hearing Level (dB HL) relative to normal hearing sensitivity (ISO, 1989). We then calculated the bilateral four tone threshold average from 88 Pre Prin ts Pre Prin ts thresholds at 500Hz, 1000Hz, 2000Hz and 4000Hz. Generally, only subjects with binaurally normal hearing (thresholds \u226420 dB HL) were included in data analysis. However, two subjects had small hearing losses in one ear only (<5 dB) and one subject had a middle ear infection in one ear. Previous unpublished research in our laboratory (and the fact that these data did not manifest as outliers), has found that isolated unilateral cases such as these do not affect end results and therefore, data from these subjects were included in analysis. speech-in-noise (sin) discrimination task : The SiN discrimination task consisted of subjects being asked to identify sentences presented in a background of multi-talker babble noise (details below). This task was administered from an HP Omnibook 4150 computer, using a program developed in-house to set noise and sentence level, to control presentation of sentences and noise, and to record, display and store results. The sentences and noise were streamed from the PC to Sennheiser HD353 headphones binaurally. Calibration of the sound stimuli was performed by coupling the headphones to a Br\u00fcel and Kj\u00e6r Artificial Ear Type 4152 containing a Br\u00fcel and Kj\u00e6r 1-inch Condenser Microphone Type 4145. The microphone output was connected to a Br\u00fcel and Kj\u00e6r Precision Sound Level Meter Type 2203 on which sound pressure levels (SPLs) were read off (using the A-weighted scale on a slow time setting). The sentence level was standardized using a reference 1kHz signal, with average RMS level set to the same value as for the sentences and stored on the computer as a .WAV file. Calibration of the background masking noise was done by playing the noise out of the headphones and again using the slow time settings to measure output level. test sentences : Test sentences came from a standard battery of clinically-used sentences (Bench, Kowal, & Bamford, 1979) adapted for Australian use (the BKB(A) list of sentences). The BKB list contains 192 sentences, each of 4-6 words of no more than two syllables. They are short, simple words and phrases imitating everyday speech and do not include questions or explanations open to interpretation. Also, these sentences contain words that have been shown to be very familiar to non-English speakers (Brouwer, et al. 2012). Each sentence 99 Pre Prin ts Pre Prin ts consists of three keywords critical for comprehension of that sentence. The sentences are pre-recorded in a female voice with an Australian accent in a neutral tone and stored as .WAV files on the computer. Sixty sentences with similar speech reception thresholds (SRTs: the signal-to-noise ratio (SNR) at which 50% of the subjects could correctly detect the sentence in background noise) were selected for use in this study. Selection and validation of these sentences have been detailed previously (Burns & Rajan, 2008; Cainer, James & Rajan, 2008; Rajan & Cainer, 2008). The sentences were randomly allocated to one of three lists classed as \u2018Low\u2019, \u2018Moderate\u2019 or \u2018High\u2019 to denote the level of the masking noise in which they were presented; sentence level was always set to 80dBA. masking noise : The masking noise was \u2018babble noise\u2019 (BN), created as described previously (Burns & Rajan, 2008; Cainer, et al., 2008; Rajan & Cainer, 2008) to give the illusion of eight voices speaking at once, known as the \u2018cocktail party\u2019 effect, digitized and stored as .WAV files. Sentences were presented to subjects in a background of one of three noise levels: 1) Low noise level at 78dBA (SNR of +2dB); 2) Moderate noise level at 81dBA (SNR of -1dB); and 3) High noise level at 84dbA (SNR of -4dB). The noise was played continuously throughout each test list and was turned off at the end of each list until just before the start of the next list. general procedures : For the SiN discrimination task each subject was instructed that they would be presented with three lists of sentences in noise, in succession. Each list would consist of 20 different sentences in a fixed background noise level of low, moderate or high. The order of lists, i.e., test SNRs was randomised between subjects except that the high noise level list was never presented first to ensure subjects did not start with the most difficult condition. The subject was asked to repeat each sentence after it was played to the best of their ability, or to indicate if they were unable to identify it at all, with no time limit imposed on giving the response. The experimenter would score the response and then play out the next sentence. After all 20 sentences in a list had been played, this 1010 Pre Prin ts Pre Prin ts procedure would be repeated twice more, with a different list of sentences and a different noise level, until all three lists had been tested. Upon confirmation that the subject understood the instructions and was ready to commence, the masking noise appropriate for the first test list was switched on and played by itself for 5s before the first sentence was played. Each sentence was scored as correct only if all three keywords were identified correctly and in correct order. Once the experimenter had scored the response, the next sentence was automatically played 1.5s later, and the test continued until all 20 sentences had been presented. Subjects were given a short break between lists. The order of presentation of sentences in each list was randomised by the software so it was unique for each subject. Scoring of performance in each list consisted of recording the percentage of sentences they were able to recall in each list. Indexing performance in the SiN task: calculating the SNR50 For data analysis, the first step was to calculate the percentage of sentences identified correctly by a subject for each list. This was done using only the middle ten sentences for each noise level for the following reasons: The first five sentences were discarded as training sentences as in our previous studies (Burns & Rajan, 2008; Cainer, et al., 2008; Rajan & Cainer, 2008), and the last five were discarded as some subjects showed signs of fatigue or loss of concentration. Then data from each subject were fitted with a linear function using regression analysis and from the regression equation the midpoint of the function \u2013 the SNR at which 50% of the sentences would be detected correctly (SNR50) was determined. These SNR50 data represented the measure derived from the SiN task as a measure of verbal working memory. We also calculated SNR50 using only the last 10 sentences of each list and found generally similar SNR50 effects. We therefore chose to use the middle 10 sentences as least likely to be affected by either training effects or loss of concentration. Academic Assessment 1111 Pre Prin ts Pre Prin ts As well as the SiN test and questionnaire, the students\u2019 academic marks were also collected from the standard academic assessments faculty databases for data analysis. This included the first and second year data for the 2008 & 2009 cohorts, but only the first year data was collated for the 2010 cohort due to time limitations. Therefore analysis for the first year results were performed using the 103 students mentioned earlier; for the second year, analysis could be performed only on 54 (from the 103) students who had completed both years of study, i.e. students from the 2008-2009 cohorts only. Course assessments varied from year to year, however all students\u2019 marks consisted of a combination of written examinations, individual coursework and objective structured clinical examination (OSCE) simulations. For data analysis nomenclature, these assessments were termed \u2018End-of-Year Totals\u2019 (Year 1 or Year 2); \u2018Coursework\u2019, comprising of essays, oral presentations and portfolios; \u2018Examinations\u2019, comprising of Multiple Choice and Short Answer Questions; and \u2018OSCEs\u2019 whereby the students undergo simulated clinical/patient scenarios at various timed stations whilst being assessed. The OSCEs were further subdivided into two categories according to the skills that were being evaluated: those in which the emphasis was primarily on technical skills (\u2018OSCE Technical\u2019, e.g., injecting techniques or taking vital signs) or those in which the emphasis was primarily on communication skills (\u2018OSCE Communications\u2019, e.g., taking a patient\u2019s history or providing an explanation to a simulated patient). structured clinical examination assessment : In Year 1, this overall model was not a strong predictor of academic achievement, but there was a significant unique contribution of SNR50 to the OSCE Communications score, indicating that vWM has a role in this assessment, and significant unique contributions of AoAoE and ELS to the OSCE Technical scores indicating that language fluency rather than vWM is involved in academic performance of the latter assessment. It is not surprising that the OSCE subcategories were the only assessments that showed significant correlations. This assessment type, particularly the Communications component, is one that has continually shown major performance differences between L1 and L2 medical students in many different countries and regardless of whether the L1 is English or another language (Fernandez, et al. 2007; 2020 Pre Prin ts Pre Prin ts Liddell & Koritsas, 2004; Schoonheim-Klein et al., 2007; Van Zanten, Boulet & McKinley, 2003; Wass, et al., 2003; Woolf, et al. 2007). We have also found similar results in a current study of a larger cohort of 872 medical students (Mann, et al., unpublished data), in which we did not measure L2 vWM or proficiency as in the present study. Our findings in this study showed that in the first year of the course, international medical students performed academically worse than their local peers in the OSCE assessment only, and not the Examinations or Coursework assessments. There were similar findings in the second year of the course; however, some groups did perform worse in all assessments including the OSCEs. The above findings of the OSCE subcategories suggest that specifically, the memorising and automated recalling of technical information may not be as challenging to vWM as the complex task of trying to express conceptual and abstract themes (i.e. higher-order cognitive processing) by the ESL students as posited by Van Merri\u00ebnboer & Sweller, (2010). Similarly, Tyler, (2001) suggests that the knowledge and familiarity of a topic will determine how well a non-native speaker will perform. Therefore, factual information that is rote-learnt, such as the OSCE Technical, will be equally easy to recall for both non-native experienced and inexperienced student doctors than unfamiliar abstract or conceptual topics, such as needed in the OSCE Communication tasks, which require good verbal working memory for the L2. Although the impairment of communication skills is more apparent in the 2nd year of study, it is important to note that we collated second year data only for the 2008 & 2009 cohorts and not for the 2010 cohort. The dynamics for the years may not be the same and each year should be examined on its\u2019 own basis. Notwithstanding, this model again predicted academic performance in the OSCE Communication assessment, suggesting both vWM and language deficits in the ESL students affect this assessment subcategory in the second year. Similarly, whilst the OSCE Technical model that was found to apply in 1st year was not overall predictive of academic achievement, there was a significant correlation of vWM for this assessment subtype in 2nd year. Together, both OSCE subcategories point to L2 vWM impairments in these 2nd year 2121 Pre Prin ts Pre Prin ts students. This may be due to the 2nd year curriculum being more difficult than basic first year outlines, and therefore the greater demands on English language skills consequently resulting in poorer performance by the ESL. This is quite possible as Collier, (1992) has stated that growth curves on normalized tests tend to flatten as students\u2019 progress in age and grade level and as the school load becomes academically more complex. Overall, our model of L2 vWM and English Language Skills was a strong predictor of academic attainment (controlling for the age English was first learnt) for the OSCE Communications assessment subcategory. The fact that the Communications assessment was the only significant model is in itself significant, as although the international students have proven English proficiency (via IELTS or TOEFL), these medical students still perform academically worse than their local counterparts in this assessment, even whilst achieving higher scores for the other subjects. Similar to the fact that we found no effects of L2 vWM on other components of assessments, in a study using L1 participants, Kidd, Watson & Gygi, (2007) found only a weak correlation between SAT scores and auditory abilities using SiN tasks. Using a broad WM test battery, Krumm, et al., (2008) also found only small indirect measures of WM as a predictor of academic performance. In contrast, Tolar, et al., (2009) found that WM strongly related to an adult\u2019s mathematical performance, but not when other cognitive factors where controlled for. Verbal WM is not the only factor poorer for an L2 learner. McDonald, (2006) reported that late English language learners had, in addition to poorer WM, poorer English decoding ability and lower speed of processing in English. Takano & Noda, (1993) posited this slower speed of L2 processing as a temporary decline in thinking ability because the demanding processing load interfered strongly with the L2 subject\u2019s thinking, beyond the normal foreign language processing difficulties experienced by non-native speakers. Takano & Noda, (1995) demonstrated that this \u201cforeign language effect\u201d was greater the more the foreign language was dissimilar to the native language, with greater 2222 Pre Prin ts Pre Prin ts performance differences between, for instance, Japanese and English than German and English, which share similar language roots. It is important to note that only 51-75% of variance in academic attainment is explained by general cognitive abilities (of which processing speed and WM are two cognitive processes) (Rohde & Thompson, 2007). It is not surprising then that correlations among working memory (or vWM) measures, e.g. reading span, generally tend to be moderate (Tolar, et al., 2009) as seen in the aforementioned studies and the results of this report.",
"v2_text": "abstract : Working memory (WM) is often poorer for a second language (L2). In low noise conditions, people listening to a language other than their first language (L1) may have similar auditory perception skills for that L2 as native listeners, but do worse in high noise conditions and this has been attributed to the poorer WM for L2. Given that WM is critical for academic success in children and young adults, these speech in noise effects have implications for academic performance where the language of instruction is L2 for a student. We used a well-established Speech-in-Noise task as a verbal WM (vWM) test and developed a model correlating vWM and measures of English proficiency and/or usage to scholastic outcomes in a multi-faceted assessment program. Our model predicted significant performance differences in an assessment requiring communications skills, but not on a companion assessment requiring knowledge of procedural skills or other assessments requiring factual knowledge. Thus, impaired vWM for an L2 appears to affect specific communications-based assessments in tertiary medical students. Keywords: Verbal Working Memory; Speech-in-Noise; Academic Attainment; OSCE; English as a Second Language; Medical 33 3 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 4 Pre Prin ts Pre Prin ts results : Speech in Noise performance and relationship to English proficiency We used the SiN task to assess the presence of vWM deficits in L2 in our medical student population. In comparing across groups, students who had learnt English as a first language (EFL) had significantly smaller SNR50 values than the students who had learnt English as a second language (ESL), (Student\u2019s-t(76) = -4.208, p<0.001) as seen in Figure 1. Twenty-five students were not included in this analysis, as they had learnt English and another language concurrently (true bilingual) and thus did not have English as a first or second language. **Insert Figure 1 here These observations established that the point of subjective performance (the SNR50) from our SiN task is a good index of verbal working memory for L2 in our medical student population. We then used correlational analysis to assess the relationship between SNR50 and English usage items from the questionnaire, as outlined in Table 2. **Insert Table 2 here Significant negative correlations were observed between seven of the nine variables and the SNR50. The remaining two variables, Perceived Stress Scale (PSS) and the Age of Acquisition of English (AoAoE), were significantly positively correlated with the SNR50, indicating that those with a higher SNR50 ratio (poorer capacity to discriminate simple English sentences from noise) had learnt English later in life, i.e. more likely the International medical students, and had higher levels of stress (as noted in the current literature). Local students exhibited significantly lower SNR50 scores than the international medical undergraduates (t(101) = 6.23, p<0.001), as well as being significantly younger when they first learnt English (t(101) = 3.33, p=0.001). No significant correlation was detected between the SNR50 and the students\u2019 Visual/Verbal Learning Style (r=-0.023), suggesting that the possible cultural variability in this factor was not a substantial confound in our findings. 1313 23 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 24 Pre Prin ts Pre Prin ts On the basis of these observations, we then conducted multiple regression analyses using the five items significantly correlated to SNR50 that pertained to English proficiency and/or usage. These variables were: Age of Acquisition of English (AoAoE); Perceived English Proficiency (PEP); how often their mother (primary caregiver) spoke English when the student was growing up (MSE); the students\u2019 own preference for speaking English (PSE); and how often the student spoke English in the last month (ESLM). All variables were entered simultaneously using the Enter method. The results showed that the variance in SNR50 was significantly predicted by this model of L2 proficiency (F(5, 93) = 9.37, p<0.001, r2=0.335), with the five variables altogether explaining 33.5% of the total variance in SNR50. There were two variables that significantly contributed to this overall variance. The first, Perceived English Proficiency (PEP), had the highest beta coefficient of -0.409 (p<0.001) and accounted for 9.8% of the variance. The other variable was MSE with a beta coefficient of -0.366 (p=0.005) and a unique contribution of 5.91% to the overall 33.5% variance. The other three variables, AoAoE, PSE and ESLM, were not significant predictors of SNR50 in this particular model with beta values of 0.020, 0.159 and -0.019 respectively. However, AoAoE and ESLM showed significant correlations with SNR50. Figure 2 graphically shows the zero-order correlations and beta coefficients for the four variables that were highly correlated to SNR50 as also shown in Table 2. **Insert Figure 2 here One caveat to interpretation of our results is that the five variables pertaining to English proficiency and usage (AoAoE, PEP, MSE, PSE and ESLM) are also highly significantly correlated with each other, with r values >0.5 (Table 2). This may suggest that these variables share the same set of underlying causal elements that affect vWM for L2 and its usage, i.e. they demonstrate multicollinearity. Therefore, a principal component analysis was performed to establish if there were underlying common constructs involved across these factors. The analysis yielded one factor with an eigenvalue >1.0 that accounted for 65% of the variance. All variables had high loadings with a 1414 25 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 26 Pre Prin ts Pre Prin ts minimum of 0.725, and a reliability test yielded a Chronbach\u2019s \u03b1 coefficient of 0.760 (considered an acceptable value of good internal consistency). In order to include all variables in this construct, it is necessary for all variables to be of the same scale. One variable, AoAoE, however, could not be changed (reverse coded) to the same scale as the other four variables in an appropriate way that did not change its correlation values. Therefore, it could not sit in this new construct and as it has been widely documented that language proficiency is influenced by the age at which the language is acquired, hierarchical analysis was conducted. The new construct of the four remaining variables, i.e. PEP, MSE, PSE and ESLM, was representative of the amount of exposure and usage the students had of English and a self-rating of their English skills. It was thus an approximation of the students\u2019 overall English proficiency, renamed \u2018English Language Skills\u2019 (ELS) and the means were calculated for analysis and checked for multicollinearity against SNR50. Hierarchical multiple regression analyses were used, controlling for AoAoE in the first step and SNR50 and the new construct ELS in the second step. Analysis was performed for the End of Year Total scores, as well as for each Assessment (as described in the Methods section) for Year 1 and Year 2 of study. Results are set out in Table 3 and discussed in detail below. **Insert Table 3 here These results establish that not only is SNR50 a good index of verbal working memory for L2, but it could be employed to test if poorer L2 vWM is a strong predictor of academic performance along with language proficiency skills. Academic performance and relationship to English language skills In the first year of study, the results showed that SNR50 and ELS were not significant predictors of overall academic performance, even when AoAoE was controlled for. However, the L2 vWM index (SNR50) did make a significant unique contribution to the OSCE Communications performance, with a beta coefficient of -0.231 (p=0.043). This demonstrated that the smaller the SNR50 1515 27 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 28 Pre Prin ts Pre Prin ts ratio (i.e., the better the vWM for discrimination of simple English sentences from noise), then the greater the Communications score. In contrast to this, results for the OSCE Technical skills showed significant positive correlations with the AoAoE (beta coefficient of 0.326, p=0.023) and with ELS (beta coefficient of 0.329, p=0.030). These correlations showed that students who had learnt English significantly later in life, but who rated their English skills more highly (International students with good English proficiency skills), performed better in the technical aspects of the OSCEs, despite learning the L2 at a later age. The SNR50 was not significant, indicating that L2 vWM does not influence academic performance for this particular assessment. Overall, after controlling for the age English was acquired, there was no clear, major predictor of academic performance in Year 1. In Year 2, this model of vWM and ELS while controlling for AoAoE was a significant predictor of academic performance of the OSCE Communications skills (p=0.008), explaining 21% of the variance of this assessment. ELS had the highest beta coefficient of 0.315 but this was not statistically significant and accounted for only 3.46% to the overall 21% variance. There was also a significant negative correlation with AoAoE on its own in Step 1 (beta coefficient = -0.384, p=0.004), but AoAoE was no longer uniquely significant in the overall model for predicting OSCE Communication skills, indicating it has only an indirect influence on predicting performance of this academic assessment. With regard to the OSCE Technical assessment for Year 2, the effects were incongruous with those observed in the results obtained for Year 1, with the SNR50 now significantly correlated (beta coefficient = 0.346, p=0.038), but AoAoE and ELS showing no correlation with academic performance. As it was the International medical students who exhibited higher SNR50 ratios, this would indicate that these students could be performing better in this category than their Local counterparts. This was confirmed by an independent samples t-test, which showed that the International medical students performed better in this assessment in Year 2 than their Local peers (t(43.73) = 3.376, p=0.002). This 1616 29 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 30 Pre Prin ts Pre Prin ts would suggest that the International students\u2019 L2 vWM is not impaired in this assessment in Year 2 (as in Year 1), perhaps because the recall of technical data is not as challenging on vWM capacity as conceptual and abstract comprehension (Van Merri\u00ebnboer & Sweller, 2010). Overall, the model is not a significant predictor for this assessment and explains only 11% of the variance, with SNR50 uniquely contributing 8.07%. Although the model was not a significant predictor of the academic performance of the 2nd year total OSCE (i.e. not subdivided into OSCE Communications and OSCE Technical), it is worth noting that it accounts for 13% of the overall variance for this variable, which in the classroom would be regarded as a considerable proportion. T-test analysis of the Year 2 OSCE scores showed that while there was no significant difference between Local and International medical students (p=0.113), there was a significant difference for the AoAoE, with students who acquired English before the age of five having better overall marks for the OSCE assessment than those who acquired English later (t(52) = 2.038, p=0.047). This is also evident in the significant negative correlation of AoAoE in Step 1, with a beta coefficient of -0.320 and significant p-value of 0.018. However, in Step 2, AoAoE was no longer significant, demonstrating that there are overlapping effects with the other variables. To summarise, after controlling for the age at which English was first learnt, verbal working memory for English (as indexed by the SNR50 in our speech-in-noise task) and ELS were not strong predictors of the overall End of Year Totals or for the individual Assessments, with the exception of the OSCEs. For the OSCE assessments, the contribution made to the variance by each predictor varied for the OSCE types and was different for each year of study. The OSCE Communications was the only significant model, which in itself is a significant finding and which is discussed later. 1717 31 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 32 Pre Prin ts Pre Prin ts discussion : The relationship between verbal Working Memory and academic attainment has been well documented in L1, particularly with young learners (Gathercole, Pickering, Knight, et al., 2004). However, the role of vWM in predicting academic achievement in L2 adults, particularly medical students, has been only occasionally examined with inconsistent effects (see Harrington & Sawyer, 1992; Juffs & Harrington, 2011). The aim of the current study was to explore if L2 vWM plays a role in academic attainment in ESL students. We indexed L2 vWM using a SiN task as a WM verbal test, as such tasks have been well documented to be a good indicator of L2 vWM and because such a task reflected, to a consistent degree, the background conditions occurring in some of the venues in which information was imparted to student doctors in their course. Linguistically, English target speech and English speech noise consist of many common properties (e.g., phonemes, syllable structures, prosodic features, etc.), which may make it more difficult for listeners, particularly non-native, to segregate target language from background noise and this may contribute to greater informational masking (Van Engen, 2010). Background masking noise can be classed as energetic or informational; energetic masking is thought to affect speech processing at the level of the auditory periphery, whereas informational masking interferes with higher-order processing such as attention and cognitive load. Informational maskers have therefore been often used in working memory tasks to good effect. Hygge et al (2003) found that meaningful irrelevant speech noise significantly impaired recall in a text-reading memory task in 92 native high school students in Sweden. We also examined a number of other factors known or postulated to influence L2 skills, in particular the age at which the participants first learnt English (as their L2) as this factor has previously been shown to influence English learning and proficiency (Johnson & Newport, 1989). In our first analysis, we confirmed that the point of subjective performance (PSE; the SNR50 score) in our SiN task was indeed a good index of verbal working memory for L2 in our student population, with our results 1818 33 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 34 Pre Prin ts Pre Prin ts showing that the EFL students had smaller PSE scores than the ESL students. This meant that the EFL medical students were better able to identify simple English words in a noisy background than the ESL medical students. This was an important step as this SiN task is free of L2 proficiency concerns that have been a major criticism of previous studies that have used measures such as the Reading Span Task (Harrington & Sawyer, 1992; Juffs & Harrington, 2011) to show differences in L2 vWM and may be one explanation for the mixed findings of past studies. We then used this index of vWM along with English Language Skills (ELS) as our model to predict academic attainment whilst controlling for the age that English was first acquired by the student (AoAoE). Different language-related factors affect different subcategories of the Objective statistical analyses : Statistical analyses were performed using SPSS v19.0.0 (SPSS Statistics Inc.) for Windows. All statistical tests were parametric, and data were checked for normality of distribution and variation. Pearson\u2019s correlation was conducted to investigate the relationship between items from the questionnaire, Perceived Stress Scale, Index of Learning Style (the visual/verbal component only was analysed as the other components are not pertinent to this particular study) and Signal to Noise Ratio (SNR50). Standard multiple regression was carried out to assess the relationship between language proficiency and verbal working memory (SNR50) and hierarchical multiple regression was used to test the ability of three measures (SNR50, age of acquisition of English and English proficiency) to predict academic performance. Student\u2019s t-tests were also used when comparing independent groups. **Insert Table 1 here. 1212 21 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 22 Pre Prin ts Pre Prin ts material and methods : participants : All participants in this study were students enrolled in the MBBS program from 2008-2010 at Monash University. This project was biphasic with all the students, in the first instance, invited to complete a questionnaire asking for information on their personal demographics, English acquisition and usage, musical abilities and two psychometric measures: Perceived Stress Scale (Cohen, 1994) and the Index of Learning Styles Questionnaire (Felder & Soloman, 1994). Stress has been found to have a negative impact on the academic performance of first year medical students, particularly international students (Bagot, et al., 2005) as well as the style of learning adopted by international versus local students, such as deep vs. surface learning styles (Newble & Entwistle, 1986). As mentioned in the Introduction, the international medical students of this course must pass stringent measures of English proficiency prior to enrolment, such as the IELTS or the TOEFL. Therefore, the questions on the survey pertained mainly to measurable English attributes such as \u2018In what order did you learn English and your other language\u2019? There was one question on the students\u2019 perceived English and Language Other Than English (LOTE) proficiency which was purely self-rated from a score of \u20180=poor\u2019 to \u20184=excellent\u2019. The surveys were distributed at the commencement of each university year in the 1st year of the medical undergraduates\u2019 course. Of the 791 questionnaires distributed over the three years, 582 were returned giving a response rate of 73.6%. Participation was voluntary and students could withdraw at any stage. In the second phase of the project, students were asked to participate in a SiN test (described below). As it was not feasible to submit all 582 subjects to this test, we performed a power analysis using GPower 3.0.10, which calculated that we would require 15 subjects in each group to give us an effect size of 0.8 at a power level of 90%. We tested a total of 113 subjects. Of these, one subject was excluded from data analysis due to hearing impairments and nine candidates were classed as outliers with means more than two standard 77 11 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 12 Pre Prin ts Pre Prin ts deviations from the sample mean (at \u03b1=0.05), leaving a total of 103 subjects, which still gave us ample power for this particular study. Analysis and findings relevant to all 582 students are currently being researched by the authors, and will be reported elsewhere; the emphasis of this report is on the outcomes of the 103 subjects undertaking the SiN test. Demographic characteristics are set out in Table 1. Students were classed as \u2018Local\u2019 if they were Australian or New Zealand citizens, or if they held permanent residency; or students were classed as \u2018International\u2019 if they held temporary entry visas, in accordance with the option chosen by the students on their questionnaires. All ethics for this study were approved by the Monash University Human Research Ethics Committee (MUHREC) project number: CF08/2667-2008001361. audiometry testing : At the outset, hearing sensitivity in each subject was measured with audiometry using a Beltone Model 110 Clinical Audiometer, calibrated to present pure tones through calibrated TDH headphones. Hearing was tested one ear at a time at 500Hz, 1000Hz, 2000Hz, 4000Hz, 6000Hz and 8000Hz. The minimum sound level at each frequency was recorded as the threshold in decibels Hearing Level (dB HL) relative to normal hearing sensitivity (ISO, 1989). We then calculated the bilateral four tone threshold average from thresholds at 500Hz, 1000Hz, 2000Hz and 4000Hz. Generally, only subjects with binaurally normal hearing (thresholds \u226420 dB HL) were included in data analysis. However, two subjects had small hearing losses in one ear only (<5 dB) and one subject had a middle ear infection in one ear. Previous unpublished research in our laboratory (and the fact that these data did not manifest as outliers), has found that isolated unilateral cases such as these do not affect end results and therefore, data from these subjects were included in analysis. speech-in-noise (sin) discrimination task : The SiN discrimination task consisted of subjects being asked to identify sentences presented in a background of multi-talker babble noise (details 88 13 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 14 Pre Prin ts Pre Prin ts below). This task was administered from an HP Omnibook 4150 computer, using a program developed in-house to set noise and sentence level, to control presentation of sentences and noise, and to record, display and store results. The sentences and noise were streamed from the PC to Sennheiser HD353 headphones binaurally. Calibration of the sound stimuli was performed by coupling the headphones to a Br\u00fcel and Kj\u00e6r Artificial Ear Type 4152 containing a Br\u00fcel and Kj\u00e6r 1-inch Condenser Microphone Type 4145. The microphone output was connected to a Br\u00fcel and Kj\u00e6r Precision Sound Level Meter Type 2203 on which sound pressure levels (SPLs) were read off (using the A-weighted scale on a slow time setting). The sentence level was standardized using a reference 1kHz signal, with average RMS level set to the same value as for the sentences and stored on the computer as a .WAV file. Calibration of the background masking noise was done by playing the noise out of the headphones and again using the slow time settings to measure output level. test sentences : Test sentences came from a standard battery of clinically-used sentences (Bench, Kowal, & Bamford, 1979) adapted for Australian use (the BKB(A) list of sentences). The BKB list contains 192 sentences, each of 4-6 words of no more than two syllables. They are short, simple words and phrases imitating everyday speech and do not include questions or explanations open to interpretation. Also, these sentences contain words that have been shown to be very familiar to non-English speakers (Brouwer, Van Engen, Calandruccio & Bradlow, 2012). Each sentence consists of three keywords critical for comprehension of that sentence. The sentences are pre-recorded in a female voice with an Australian accent in a neutral tone and stored as .WAV files on the computer. Sixty sentences with similar speech reception thresholds (SRTs: the signal-to-noise ratio (SNR) at which 50% of the subjects could correctly detect the sentence in background noise) were selected for use in this study. Selection and validation of these sentences have been detailed previously (Burns & Rajan, 2008; Cainer, James & Rajan, 2008; Rajan & Cainer, 2008). The sentences were randomly allocated to one of three lists classed as \u2018Low\u2019, 99 15 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 16 Pre Prin ts Pre Prin ts \u2018Moderate\u2019 or \u2018High\u2019 to denote the level of the masking noise in which they were presented; sentence level was always set to 80dBA. masking noise : The masking noise was \u2018babble noise\u2019 (BN), created as described previously (Burns & Rajan, 2008; Cainer, et al., 2008; Rajan & Cainer, 2008) to give the illusion of eight voices speaking at once, known as the \u2018cocktail party\u2019 effect, digitized and stored as .WAV files. Sentences were presented to subjects in a background of one of three noise levels: 1) Low noise level at 78dBA (SNR of +2dB); 2) Moderate noise level at 81dBA (SNR of -1dB); and 3) High noise level at 84dbA (SNR of -4dB). The noise was played continuously throughout each test list and was turned off at the end of each list until just before the start of the next list. general procedures : For the SiN discrimination task each subject was instructed that they would be presented with three lists of sentences in noise, in succession. Each list would consist of 20 different sentences in a fixed background noise level of low, moderate or high. The order of lists, i.e., test SNRs was randomised between subjects except that the high noise level list was never presented first to ensure subjects did not start with the most difficult condition. The subject was asked to repeat each sentence after it was played to the best of their ability, or to indicate if they were unable to identify it at all, with no time limit imposed on giving the response. The experimenter would score the response and then play out the next sentence. After all 20 sentences in a list had been played, this procedure would be repeated twice more, with a different list of sentences and a different noise level, until all three lists had been tested. Upon confirmation that the subject understood the instructions and was ready to commence, the masking noise appropriate for the first test list was switched on and played by itself for 5s before the first sentence was played. Each sentence was scored as correct only if all three keywords were identified correctly and in correct order. Once the experimenter had scored the response, the next sentence was automatically played 1.5s later, and the test continued until all 20 sentences had been presented. Subjects were given a short break 1010 17 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 18 Pre Prin ts Pre Prin ts between lists. The order of presentation of sentences in each list was randomised by the software so it was unique for each subject. Scoring of performance in each list consisted of recording the percentage of sentences they were able to recall in each list. Indexing performance in the SiN task: calculating the SNR50 For data analysis, the first step was to calculate the percentage of sentences identified correctly by a subject for each list. This was done using only the middle ten sentences for each noise level for the following reasons: The first five sentences were discarded as training sentences as in our previous studies (Burns & Rajan, 2008; Cainer, et al., 2008; Rajan & Cainer, 2008), and the last five were discarded as some subjects showed signs of fatigue or loss of concentration. Then data from each subject was fitted with a linear function using regression analysis and from the regression equation the midpoint of the function \u2013 the SNR at which 50% of the sentences would be detected correctly (SNR50) was determined. These SNR50 data represented the measure derived from the SiN task as a measure of verbal working memory. We also calculated SNR50 using only the last 10 sentences of each list and found generally similar SNR50 effects. We therefore chose to use the middle 10 sentences as least likely to be affected by either training effects or loss of concentration. academic assessment : As well as the SiN test and questionnaire, the students\u2019 academic marks were also collected for data analysis. This included the first and second year data for the 2008 & 2009 cohorts, but only the first year data was collated for the 2010 cohort due to time limitations. Course assessments varied from year to year, however all students\u2019 marks consisted of a combination of written examinations, individual coursework and objective structured clinical examination (OSCE) simulations. For data analysis nomenclature, these assessments were termed \u2018End-of-Year Totals\u2019 (Year 1 or Year 2); \u2018Coursework\u2019, comprising of essays, oral presentations and portfolios; \u2018Examinations\u2019, comprising of Multiple Choice and 1111 19 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 20 Pre Prin ts Pre Prin ts Short Answer Questions; and \u2018OSCEs\u2019 whereby the students undergo simulated clinical/patient scenarios at various timed stations whilst being assessed. The OSCEs were further subdivided into two categories according to the skills that were being evaluated: those in which the emphasis was primarily on technical skills (\u2018OSCE Technical\u2019, e.g., injecting techniques or taking vital signs) or those in which the emphasis was primarily on communication skills (\u2018OSCE Communications\u2019, e.g., taking a patient\u2019s history or providing an explanation to a simulated patient). structured clinical examination assessment : In Year 1, this overall model was not a strong predictor of academic achievement, but there was a significant unique contribution of SNR50 to the OSCE Communications score, indicating that vWM has a role in this assessment, and significant unique contributions of AoAoE and ELS to the OSCE Technical scores indicating that language fluency rather than vWM is involved in academic performance of the latter assessment. It is not surprising that the OSCE subcategories were the only assessments that showed significant correlations. This assessment type, particularly the Communications component, is one that has continually shown major performance differences between L1 and L2 medical students in many different countries and regardless of whether the L1 is English or another language (Fernandez, Wang, Braveman, Finkas & Hauer, 2007; Liddell & Koritsas, 2004; Schoonheim-Klein et al., 2007; Van Zanten, Boulet & McKinley, 2003; Wass, et al., 2003; Woolf, Haq, McManus, Higham & Dacre, 2007). We have also found similar results in a current study of a larger cohort of 872 medical students (Mann, Canny, Lindley & Rajan, unpublished), in which we did not measure L2 vWM or proficiency as in the present study. Our findings in this study showed that in the first year of the course, international medical students performed academically worse than their local peers in the OSCE 1919 35 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 36 Pre Prin ts Pre Prin ts assessment only, and not the Examinations or Coursework assessments. There were similar findings in the second year of the course; however, some groups did perform worse in all assessments including the OSCEs. The above findings of the OSCE subcategories suggest that specifically, the memorising and automated recalling of technical information may not be as challenging to vWM as the complex task of trying to express conceptual and abstract themes (i.e. higher-order cognitive processing) by the ESL students as posited by Van Merri\u00ebnboer and Sweller (2010). Similarly, Tyler (2001) suggests that the knowledge and familiarity of a topic will determine how well a non-native speaker will perform. Therefore, factual information that is rote-learnt, such as the OSCE Technical, will be equally easy to recall for both non-native experienced and inexperienced student doctors than unfamiliar abstract or conceptual topics, such as needed in the OSCE Communication tasks, which require good verbal working memory for the L2. Although the impairment of communication skills is more apparent in the 2nd year of study, it is important to note that we collated second year data only for the 2008 & 2009 cohorts and not for the 2010 cohort. The dynamics for the years may not be the same and each year should be examined on its\u2019 own basis. Notwithstanding, this model again predicted academic performance in the OSCE Communication assessment, suggesting both vWM and language deficits in the ESL students affect this assessment subcategory in the second year. Similarly, whilst the OSCE Technical model that was found to apply in 1st year was not overall predictive of academic achievement, there was a significant correlation of vWM for this assessment subtype in 2nd year. Together, both OSCE subcategories point to L2 vWM impairments in these 2nd year students. This may be due to the 2nd year curriculum being more difficult than basic first year outlines, and therefore the greater demands on English language skills consequently resulting in poorer performance by the ESL. This is quite possible as Collier (1992) has stated that growth curves on normalized tests tend to flatten as students\u2019 progress in age and grade level and as the school load becomes academically more complex. 2020 37 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 38 Pre Prin ts Pre Prin ts Overall, our model of L2 vWM (as indexed by the students\u2019 PSE in the SiN task) and English Language Skills was a strong predictor of academic attainment (controlling for the age English was first learnt) for the OSCE Communications assessment subcategory. The fact that the Communications assessment was the only significant model is in itself significant, as although the international students have proven English proficiency (via IELTS or TOEFL), these medical students still perform academically worse than their local counterparts in this assessment, even whilst achieving higher scores for the other subjects. Similar to the fact that we found no effects of L2 vWM on other components of assessments, in a study using L1 participants, Kidd, Watson & Gygi (2007) found only a weak correlation between SAT scores and auditory abilities using SiN tasks. Using a broad WM test battery, Krumm, et al., (2008) also found only small indirect measures of WM as a predictor of academic performance. In contrast, Tolar, et al., (2009) found that WM strongly related to an adult\u2019s mathematical performance, but not when other cognitive factors where controlled for. Verbal WM is not the only factor poorer for an L2 learner. McDonald (2006) reported that late English language learners had, in addition to poorer WM, poorer English decoding ability and lower speed of processing in English. Takano and Noda (1993) posited this slower speed of L2 processing as a temporary decline in thinking ability because the demanding processing load interfered strongly with the L2 subject\u2019s thinking, beyond the normal foreign language processing difficulties experienced by non-native speakers. Takano & Noda (1995) demonstrated that this \u201cforeign language effect\u201d was greater the more the foreign language was dissimilar to the native language, with greater performance differences between, for instance, Japanese and English than German and English, which share similar language roots. It is important to note that only 51-75% of variance in academic attainment is explained by general cognitive abilities (of which processing speed and WM are two cognitive processes) (Rohde & Thompson, 2007). It is not surprising then that correlations among working memory (or vWM) measures, 2121 39 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 40 Pre Prin ts Pre Prin ts e.g. reading span, generally tend to be moderate (Tolar, et al., 2009) as seen in the aforementioned studies and the results of this report. conclusions : In summary, our study contributes to the growing research examining why non-native medical undergraduates generally perform academically worse than their native speaker counterparts. The implications are that in a prestigious course such as the MBBS degree, where all students have proven high academic abilities, motivation and expectations prior to commencement, small differences at the early stages could have disproportionate impacts on the medical careers of L2 students, for example, in selection for highly competitive specialist training positions or fellowships. Finally, it highlights an area where international medical students continually to fall down despite rigorous processes and comparable English proficiency. 2222 41 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 42 Pre Prin ts Pre Prin ts",
"url": "https://peerj.com/articles/23/reviews/",
"review_1": "Valeria Souza \u00b7 Jan 9, 2013 \u00b7 Academic Editor\nACCEPT\nThanks very much for promply making all the suggested corrections, I think the paper is really interesting and should be published.",
"review_2": "Valeria Souza \u00b7 Dec 24, 2012 \u00b7 Academic Editor\nMINOR REVISIONS\nThis paper describes bacterial communities that can use methane derived from sediment in Lake Washington and examined in the context of 4 microcosms. The authors provide evidence for potential cooperation between two methane using bacteria in a coordinated response to methane and nitrate. They have conducted careful experiments and obtained a nice modern data set, using 13C-methane in 4 microcosms with different conditions and both 16S and shotgun WGS of all 4 microcosms in addition to the sediment itself. The use of SIP isolates the organisms involved in the process they are interested in, a very effective and still somewhat unusual approach. While their excel-generated description of the alpha diversity does enable some aspects of the data to be described, this is too much data for the eye to compare entirely in terms of abundance in tables and pie charts. The microbial ecology community has developed user friendly tools for this type of analysis which would greatly enhance the readers ability to understand the conclusions presented here, and may also help the authors understand their data better and most of all, help the reader.",
"review_3": "Reviewer 1 \u00b7 Dec 23, 2012\nBasic reporting\nNo Comments\nExperimental design\nNo Comments\nValidity of the findings\nNo Comments\nAdditional comments\nMajor\n\nThe figures do not capture some of the most interesting aspects of their data, including the 13C methane experiment timeline and the beta diversity comparing the microcosms, differences between 16S and WGS metagenome sequence data. I suggest the use of MG-Rast, mother and/or Qiime for the 16S data, ideally followed by ecology ordination analysis such as MDS using a Bray-Curtis distance matrix from the abundance tables (this can be done in R, see the vegan package and others, Primer, Qiime and many other software).\n\nThroughout the text, when the taxonomy of hits to a particular gene are shown, they are referred to as a phylogenetic profile. My understanding of a phylogenetic profile is of an analysis of the patterns of co-occurance of multiple proteins across phyla, which sometimes helps t annotate unknown proteins that may be part of the same pathway. A phylogenetic profile of the sequenced methane-utlizing genomes described in this manuscript and compared with what is found in the metagenomes does sound interesting. The approach taken by Eisen and colleagues using freely available software designed for microarray analysis might be useful (see Fig 7 from PLoS Genet. 2005 Nov;1(5):e65. Epub 2005 Nov 25.)\n\nOtherwise, please replace the term \"phylogenetic profile\" with \"taxonomy of\" when describing the taxonomy of genes found in the metagenomes.\n\n\nSuggestions for figures:\n\n- Consider replacing Pie charts with bar or lattice charts (google \"Why pie charts are bad\")\n- add a figure with a schematic describing the microcosms and and experimental design\n\n- A figure with days to heavy band on the X axes and microcosm on the Y axes\n- Figure 1 in the form of bar charts and with the addition of MDS or unifrac or another clustering technique would be much easier to understand\n- MDS or other analysis comparing pyrotag and WGS sequencing\n\n\nMinor\n\n- The figures need labels in the figure itself rather than only in the legend. Rather than giving details about those legends, re-structuring as described above should hopefully lead to legends that contain a single important message rather than catalog-like descriptions of large amounts of data.\n\n- Labelling sections in the text with headings describing what is in the section would also help guide the reader\n\n- Figure 3: it would be helpful to understand the overall abundance of what we are looking at here. In 15-20, are there few reads attributed here overall, as they are mostly composed of elements that are described as rare overall? A legend (labels for the colors in the figure itself) would also be very helpful here. The first line of the figure legend here and for all figures should summarize an important result from the figure rather than just describing an encyclopedic catalog\n\nline 105: genes annotated as such > identified as\n\nline 110: do you mean all genes from those organisms, or just the sequence for pmoB and mmoX?\n\nline 210: being matched > matching\nline 212 respectively) the > respectively). The\nline 276 missing period after Figures 1,2)\n\nTable 1: add read length and # of reads\nmany table legends: remind readers what function the abbreviated gene names encode\nCite this review as\nAnonymous Reviewer (2013) Peer Review #1 of \"A metagenomic insight into freshwater methane-utilizing communities and evidence for cooperation between the Methylococcaceae and the Methylophilaceae (v0.1)\". PeerJ https://doi.org/10.7287/peerj.23v0.1/reviews/1",
"review_4": "Lisa Stein \u00b7 Dec 19, 2012\nBasic reporting\nno comments\nExperimental design\nno comments\nValidity of the findings\nno comments\nAdditional comments\nThis is a well executed and well written study. Minor suggestions for improvement are:\n1) watch for consistent spelling of genus names throughout\n2) since there was a such a major difference in the number of reads between libraries, the authors might want to comment on how the data were normalized in order to make proper comparisons in percentages of certain gene categories between experiments.\n3) in the conclusion, mention which groups of methanotrophs were likely cooperating with the methylotrophs\n4) since nitrate had a strong effect on the populations, are the authors suggesting that it is stimulating the community as an N-source? If so, then perhaps instead of ending the paper with the conclusion that the population is likely not carrying out respiratory denitrification, they could conclude with the possibility that nitrate is stimulating growth as a nutrient.\nCite this review as\nStein L (2013) Peer Review #2 of \"A metagenomic insight into freshwater methane-utilizing communities and evidence for cooperation between the Methylococcaceae and the Methylophilaceae (v0.1)\". PeerJ https://doi.org/10.7287/peerj.23v0.1/reviews/2",
"pdf_1": "https://peerj.com/articles/23v0.2/submission",
"pdf_2": "https://peerj.com/articles/23v0.1/submission",
"review_5": "Luis David Alcaraz \u00b7 Dec 11, 2012\nBasic reporting\nI enjoyed reading Beck et al. Paper. The authors wanted to show three things: Determining guilds involved in methane oxidation in freshwater lakes; 2. The role of Methylophilaceae in methane oxidation and; 3. The effect of nitrate on methane-oxidizing communities. They sampled a 63 m depth sample of sediment and used WGS metagenomics and amplicons of 16S rRNA for species diversity profiles, as well as some genes involved in methane oxidation as phylogenetic markers. They perform microcosm experiments enriching the atmosfere with isotope labeled 13CH4 and playing with different concentrations of Nitrogen and Oxigen to perform the experiments and measure the communities changes across time of incubation under this conditions.\n\nThe paper contributes with understanding on methane oxidating species diversity and its changes across the time, assuming that the isotope marked 13C is incorporated into its DNA. However the experimental design can not help to distinguish the atmosphere enrichment from the natural death/decay of the community members across time. So the paper needs to be rephrased to make clear this point. And specifically to remove the statement that the methane-oxidizing species are product of the enrichment, that could be, but the experimental design doesn't help to conclude this. This concern is detailed in Experimental Design section of the report.\n\nThink that the authors' expertise will answer/solve the particular observations in a short time and would be enough to publish the article.\nExperimental design\nMy main concern about the design is that the authors are enriching the microcosm atmosphere (4 treatments) with very different times of incubation (10, 15, 20, and 30 days), and then use time 0 to compare the community structure. The concern here is that not being able to differentiate between the effect of the atmosphere enrichment or the aging/dead of the least resilient members of the community. I would suggest for further studies to have controls for each one of the incubation times, unamended samples of the same age for each incubation time to differ between treatments.\nValidity of the findings\nThe main findings of the paper will remain, but discussion and conclusions need to be softened due to the experimental design. The authors asses the community structure by different approaches and this is solid. Although some remarks on how to select the thresholds is required and needs to be detailed in the methods section of the paper.\nAdditional comments\nFor the nitrogen metabolism genes relying on the annotation (p5 L122-123) only of the NCBI's (nt) makes me think about the huge amount of possible false positives lying on database due to annotation errors spreading. Would suggest to use a curated database and then search through profiles or position algorithms like hmmer or psi-blast to precisely annotate your dataset.\n\nP6 L137-156 This section should fit better into methods of the paper, not results.\n\nP7 L159-162. Please move this paragraph into methods.\n\nP7 L163;171 The clusters are made at a 97% sequence identity, this is also known as OTUs, in this results you are only showing the clustering numbers, but neither here or in any table I looked into founded the total amount of sequences without clustering. This helps to see if you have some bias due to uneven coverage of each of the samples. Please incorporate the raw pyrotag sequences numbers. It would be helpful too to use an estimator of species abundance that does not rely on sequencing depth like the non-parametric estimators CHAO1 or ACE.\n\nP7 L 171;173 This is one of the phrases not supported by the design: pleas rephrase from increased, decreased from one condition into other.\n\nEven though the authors count with 5 WGS metagenomes the soul of the paper relies on phylogenetic profiling most of it with single gene approaches. The genes are obtained via PCR and then pyrotag sequencing or taxonomically assigned through an identity percentage. I think it would help to the reader to include precisely the rational behind the thresholds used in this study (i.e. p5. L114-116). P8 L199-201 Please include references to the thresholds here. Are you using Konstantinidis & Tiedje cut-offs?\n\nGoris, J., Konstantinidis, K. T., Klappenbach, J. a, Coenye, T., Vandamme, P., & Tiedje, J. M. (2007). DNA-DNA hybridization values and their relationship to whole-genome sequence similarities. International journal of systematic and evolutionary microbiology, 57(Pt 1), 81\u201391.\n\nKonstantinidis, K. T., & Tiedje, J. M. (2005). Genomic insights that advance the species definition for prokaryotes. Proceedings of the National Academy of Sciences of the United States of America, 102(7), 2567\u201372.\n\nHave you considered to use whole genome recruitment of the metagenomes? I think it could help a lot to the paper, with figures to each reference genome, and think would not be time consuming like here:\n\nBelda-Ferre P, Cabrera-Rubio R, Moya A, Mira A (2011) Mining Virulence Genes Using Metagenomics. PLoS ONE 6(10)\n\np10, L 233 consider use bin rather than categorized\n\np12, L287 what is a significant sequence sampling?\n\np12, L298 were most numerous. How most numerous than? Please indicate the raw numbers.\n\np12, L303 also low. The whole sentence does not make sense without numbers.\n\nP12, L306 most diverse.\n\np.13, L314 significant diversity = ? Please try to get diversity indexes (Simpson, Shannon, CHAO1, ACE).\n\np. 15, L384 most abundant = ?\n\nThrough out the text, be careful with the use of response, because it is not possible to say that this is a direct effect of the enrichments.\n\nTable 1. Add the raw count of pyrotag sequences, and at least one diversity index across the samples.\n\nTable 2. How is it possible to have more sequences assigned at the 90% than at the 60% cut-off? See row 1 of Methylobacter, Methylocystis, and Methylotenera\n\nTable 4, 5, I would suggest to use taxonomy profile/distribution rather than diversity.\n\nTable 6 And total number of OTUs?\n\nFigure 1. Some tags in the figure would help a lot to understand, just a bracket for A,B that states Pyrotags. A second bracket for C, D that states metagenomes\n\nFigure 3. Please change the numbers from X-axis use symbols rather than numbers, and use a key to identify each treatment. * + % # - or whatever sign you choose rather than numbers from 1 \u2013 5 will help the reader a lot to get into the figure.\nCite this review as\nAlcaraz LD (2013) Peer Review #3 of \"A metagenomic insight into freshwater methane-utilizing communities and evidence for cooperation between the Methylococcaceae and the Methylophilaceae (v0.1)\". PeerJ https://doi.org/10.7287/peerj.23v0.1/reviews/3",
"all_reviews": "Review 1: Valeria Souza \u00b7 Jan 9, 2013 \u00b7 Academic Editor\nACCEPT\nThanks very much for promply making all the suggested corrections, I think the paper is really interesting and should be published.\nReview 2: Valeria Souza \u00b7 Dec 24, 2012 \u00b7 Academic Editor\nMINOR REVISIONS\nThis paper describes bacterial communities that can use methane derived from sediment in Lake Washington and examined in the context of 4 microcosms. The authors provide evidence for potential cooperation between two methane using bacteria in a coordinated response to methane and nitrate. They have conducted careful experiments and obtained a nice modern data set, using 13C-methane in 4 microcosms with different conditions and both 16S and shotgun WGS of all 4 microcosms in addition to the sediment itself. The use of SIP isolates the organisms involved in the process they are interested in, a very effective and still somewhat unusual approach. While their excel-generated description of the alpha diversity does enable some aspects of the data to be described, this is too much data for the eye to compare entirely in terms of abundance in tables and pie charts. The microbial ecology community has developed user friendly tools for this type of analysis which would greatly enhance the readers ability to understand the conclusions presented here, and may also help the authors understand their data better and most of all, help the reader.\nReview 3: Reviewer 1 \u00b7 Dec 23, 2012\nBasic reporting\nNo Comments\nExperimental design\nNo Comments\nValidity of the findings\nNo Comments\nAdditional comments\nMajor\n\nThe figures do not capture some of the most interesting aspects of their data, including the 13C methane experiment timeline and the beta diversity comparing the microcosms, differences between 16S and WGS metagenome sequence data. I suggest the use of MG-Rast, mother and/or Qiime for the 16S data, ideally followed by ecology ordination analysis such as MDS using a Bray-Curtis distance matrix from the abundance tables (this can be done in R, see the vegan package and others, Primer, Qiime and many other software).\n\nThroughout the text, when the taxonomy of hits to a particular gene are shown, they are referred to as a phylogenetic profile. My understanding of a phylogenetic profile is of an analysis of the patterns of co-occurance of multiple proteins across phyla, which sometimes helps t annotate unknown proteins that may be part of the same pathway. A phylogenetic profile of the sequenced methane-utlizing genomes described in this manuscript and compared with what is found in the metagenomes does sound interesting. The approach taken by Eisen and colleagues using freely available software designed for microarray analysis might be useful (see Fig 7 from PLoS Genet. 2005 Nov;1(5):e65. Epub 2005 Nov 25.)\n\nOtherwise, please replace the term \"phylogenetic profile\" with \"taxonomy of\" when describing the taxonomy of genes found in the metagenomes.\n\n\nSuggestions for figures:\n\n- Consider replacing Pie charts with bar or lattice charts (google \"Why pie charts are bad\")\n- add a figure with a schematic describing the microcosms and and experimental design\n\n- A figure with days to heavy band on the X axes and microcosm on the Y axes\n- Figure 1 in the form of bar charts and with the addition of MDS or unifrac or another clustering technique would be much easier to understand\n- MDS or other analysis comparing pyrotag and WGS sequencing\n\n\nMinor\n\n- The figures need labels in the figure itself rather than only in the legend. Rather than giving details about those legends, re-structuring as described above should hopefully lead to legends that contain a single important message rather than catalog-like descriptions of large amounts of data.\n\n- Labelling sections in the text with headings describing what is in the section would also help guide the reader\n\n- Figure 3: it would be helpful to understand the overall abundance of what we are looking at here. In 15-20, are there few reads attributed here overall, as they are mostly composed of elements that are described as rare overall? A legend (labels for the colors in the figure itself) would also be very helpful here. The first line of the figure legend here and for all figures should summarize an important result from the figure rather than just describing an encyclopedic catalog\n\nline 105: genes annotated as such > identified as\n\nline 110: do you mean all genes from those organisms, or just the sequence for pmoB and mmoX?\n\nline 210: being matched > matching\nline 212 respectively) the > respectively). The\nline 276 missing period after Figures 1,2)\n\nTable 1: add read length and # of reads\nmany table legends: remind readers what function the abbreviated gene names encode\nCite this review as\nAnonymous Reviewer (2013) Peer Review #1 of \"A metagenomic insight into freshwater methane-utilizing communities and evidence for cooperation between the Methylococcaceae and the Methylophilaceae (v0.1)\". PeerJ https://doi.org/10.7287/peerj.23v0.1/reviews/1\nReview 4: Lisa Stein \u00b7 Dec 19, 2012\nBasic reporting\nno comments\nExperimental design\nno comments\nValidity of the findings\nno comments\nAdditional comments\nThis is a well executed and well written study. Minor suggestions for improvement are:\n1) watch for consistent spelling of genus names throughout\n2) since there was a such a major difference in the number of reads between libraries, the authors might want to comment on how the data were normalized in order to make proper comparisons in percentages of certain gene categories between experiments.\n3) in the conclusion, mention which groups of methanotrophs were likely cooperating with the methylotrophs\n4) since nitrate had a strong effect on the populations, are the authors suggesting that it is stimulating the community as an N-source? If so, then perhaps instead of ending the paper with the conclusion that the population is likely not carrying out respiratory denitrification, they could conclude with the possibility that nitrate is stimulating growth as a nutrient.\nCite this review as\nStein L (2013) Peer Review #2 of \"A metagenomic insight into freshwater methane-utilizing communities and evidence for cooperation between the Methylococcaceae and the Methylophilaceae (v0.1)\". PeerJ https://doi.org/10.7287/peerj.23v0.1/reviews/2\nReview 5: Luis David Alcaraz \u00b7 Dec 11, 2012\nBasic reporting\nI enjoyed reading Beck et al. Paper. The authors wanted to show three things: Determining guilds involved in methane oxidation in freshwater lakes; 2. The role of Methylophilaceae in methane oxidation and; 3. The effect of nitrate on methane-oxidizing communities. They sampled a 63 m depth sample of sediment and used WGS metagenomics and amplicons of 16S rRNA for species diversity profiles, as well as some genes involved in methane oxidation as phylogenetic markers. They perform microcosm experiments enriching the atmosfere with isotope labeled 13CH4 and playing with different concentrations of Nitrogen and Oxigen to perform the experiments and measure the communities changes across time of incubation under this conditions.\n\nThe paper contributes with understanding on methane oxidating species diversity and its changes across the time, assuming that the isotope marked 13C is incorporated into its DNA. However the experimental design can not help to distinguish the atmosphere enrichment from the natural death/decay of the community members across time. So the paper needs to be rephrased to make clear this point. And specifically to remove the statement that the methane-oxidizing species are product of the enrichment, that could be, but the experimental design doesn't help to conclude this. This concern is detailed in Experimental Design section of the report.\n\nThink that the authors' expertise will answer/solve the particular observations in a short time and would be enough to publish the article.\nExperimental design\nMy main concern about the design is that the authors are enriching the microcosm atmosphere (4 treatments) with very different times of incubation (10, 15, 20, and 30 days), and then use time 0 to compare the community structure. The concern here is that not being able to differentiate between the effect of the atmosphere enrichment or the aging/dead of the least resilient members of the community. I would suggest for further studies to have controls for each one of the incubation times, unamended samples of the same age for each incubation time to differ between treatments.\nValidity of the findings\nThe main findings of the paper will remain, but discussion and conclusions need to be softened due to the experimental design. The authors asses the community structure by different approaches and this is solid. Although some remarks on how to select the thresholds is required and needs to be detailed in the methods section of the paper.\nAdditional comments\nFor the nitrogen metabolism genes relying on the annotation (p5 L122-123) only of the NCBI's (nt) makes me think about the huge amount of possible false positives lying on database due to annotation errors spreading. Would suggest to use a curated database and then search through profiles or position algorithms like hmmer or psi-blast to precisely annotate your dataset.\n\nP6 L137-156 This section should fit better into methods of the paper, not results.\n\nP7 L159-162. Please move this paragraph into methods.\n\nP7 L163;171 The clusters are made at a 97% sequence identity, this is also known as OTUs, in this results you are only showing the clustering numbers, but neither here or in any table I looked into founded the total amount of sequences without clustering. This helps to see if you have some bias due to uneven coverage of each of the samples. Please incorporate the raw pyrotag sequences numbers. It would be helpful too to use an estimator of species abundance that does not rely on sequencing depth like the non-parametric estimators CHAO1 or ACE.\n\nP7 L 171;173 This is one of the phrases not supported by the design: pleas rephrase from increased, decreased from one condition into other.\n\nEven though the authors count with 5 WGS metagenomes the soul of the paper relies on phylogenetic profiling most of it with single gene approaches. The genes are obtained via PCR and then pyrotag sequencing or taxonomically assigned through an identity percentage. I think it would help to the reader to include precisely the rational behind the thresholds used in this study (i.e. p5. L114-116). P8 L199-201 Please include references to the thresholds here. Are you using Konstantinidis & Tiedje cut-offs?\n\nGoris, J., Konstantinidis, K. T., Klappenbach, J. a, Coenye, T., Vandamme, P., & Tiedje, J. M. (2007). DNA-DNA hybridization values and their relationship to whole-genome sequence similarities. International journal of systematic and evolutionary microbiology, 57(Pt 1), 81\u201391.\n\nKonstantinidis, K. T., & Tiedje, J. M. (2005). Genomic insights that advance the species definition for prokaryotes. Proceedings of the National Academy of Sciences of the United States of America, 102(7), 2567\u201372.\n\nHave you considered to use whole genome recruitment of the metagenomes? I think it could help a lot to the paper, with figures to each reference genome, and think would not be time consuming like here:\n\nBelda-Ferre P, Cabrera-Rubio R, Moya A, Mira A (2011) Mining Virulence Genes Using Metagenomics. PLoS ONE 6(10)\n\np10, L 233 consider use bin rather than categorized\n\np12, L287 what is a significant sequence sampling?\n\np12, L298 were most numerous. How most numerous than? Please indicate the raw numbers.\n\np12, L303 also low. The whole sentence does not make sense without numbers.\n\nP12, L306 most diverse.\n\np.13, L314 significant diversity = ? Please try to get diversity indexes (Simpson, Shannon, CHAO1, ACE).\n\np. 15, L384 most abundant = ?\n\nThrough out the text, be careful with the use of response, because it is not possible to say that this is a direct effect of the enrichments.\n\nTable 1. Add the raw count of pyrotag sequences, and at least one diversity index across the samples.\n\nTable 2. How is it possible to have more sequences assigned at the 90% than at the 60% cut-off? See row 1 of Methylobacter, Methylocystis, and Methylotenera\n\nTable 4, 5, I would suggest to use taxonomy profile/distribution rather than diversity.\n\nTable 6 And total number of OTUs?\n\nFigure 1. Some tags in the figure would help a lot to understand, just a bracket for A,B that states Pyrotags. A second bracket for C, D that states metagenomes\n\nFigure 3. Please change the numbers from X-axis use symbols rather than numbers, and use a key to identify each treatment. * + % # - or whatever sign you choose rather than numbers from 1 \u2013 5 will help the reader a lot to get into the figure.\nCite this review as\nAlcaraz LD (2013) Peer Review #3 of \"A metagenomic insight into freshwater methane-utilizing communities and evidence for cooperation between the Methylococcaceae and the Methylophilaceae (v0.1)\". PeerJ https://doi.org/10.7287/peerj.23v0.1/reviews/3\nReview 6: \n",
"LLM_extracted_review": "['Review 1: \\n- The paper is interesting and should be published.\\n\\nReview 2: \\n- The paper describes bacterial communities using methane from Lake Washington in 4 microcosms.\\n- Evidence for cooperation between methane-using bacteria is provided.\\n- The experiments are careful and the data set is modern, using 13C-methane.\\n- The alpha diversity description is too complex for visual comparison; user-friendly tools for analysis are recommended.\\n- Figures do not capture interesting data aspects; suggestions for using specific software for analysis are made.\\n- The term \"phylogenetic profile\" is misused; it should be replaced with \"taxonomy of.\"\\n- Suggestions for improving figures include replacing pie charts with bar charts and adding schematic diagrams.\\n- Figures need labels within the figure itself, and section headings in the text would help guide the reader.\\n- Clarifications are needed in the text regarding gene annotations and overall abundance in figures.\\n- Table 1 should include read length and number of reads; legends should remind readers of gene functions.\\n\\nReview 3: \\n- No comments on basic reporting, experimental design, or validity of findings.\\n- Figures do not adequately capture interesting data aspects; suggestions for software and analysis methods are provided.\\n- The term \"phylogenetic profile\" should be replaced with \"taxonomy of.\"\\n- Suggestions for figure improvements include clearer labeling and restructuring legends for clarity.\\n\\nReview 4: \\n- The study is well executed and well written.\\n- Consistent spelling of genus names should be monitored.\\n- The authors should comment on data normalization due to differences in read numbers.\\n- The conclusion should specify which methanotroph groups are cooperating with methylotrophs.\\n- The role of nitrate as a potential nutrient source should be discussed in the conclusion.\\n\\nReview 5: \\n- The paper aims to determine guilds involved in methane oxidation, the role of Methylophilaceae, and the effect of nitrate on communities.\\n- The experimental design does not distinguish between atmospheric enrichment and natural decay of community members.\\n- The authors should clarify that methane-oxidizing species may not solely result from enrichment.\\n- Concerns about the experimental design regarding incubation times and controls are raised.\\n- Discussion and conclusions need to be softened due to the design limitations.\\n- Suggestions for using curated databases for gene annotation and including raw sequence numbers are made.\\n- Recommendations for improving clarity in figures and tables are provided, including the use of symbols instead of numbers for treatment identification.']"
} |