1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Bilingualism: Revisiting Theoretical and Methodological Issues

Running head: Methodological Issues in Bilingualism Tanya Dash, Centre of Behavioural and Cognitive Sciences, University of Allahabad, Allahabad, UP, India. Bhoomika R. Kar* Centre of Behavioural and Cognitive Sciences, University of Allahabad, Allahabad, UP, India.

Correspondence: Dr. Bhoomika R Kar Centre of Behavioural and Cognitive Sciences Senate hall campus University of Allahabad Allahabad-211002 Uttar Pradesh, India bhoomika2000@yahoo.com bhoomika@cbcs.ac.in

21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49

Abstract Purpose: In the present study we examined the nature of L1 and L2 proficiency looking at relationship between objective and subjective measures of language proficiency among bilingual individuals. The current paper also intends to explore the appropriate measures of language proficiency. Method: An indigenous tool for testing language proficiency in Hindi and English was administered to 85 bilingual adults. Results: Results demonstrate better performance on L1 in all four domains of language function as compared to L2. The level of performance on speaking/ understanding tasks was better than reading/writing tasks. There was no correlation between self reported information on age of acquisition and objective measures of language proficiency. Total scores in both the domains and in both L1 and L2 showed normal distribution. Self reported information on L2 correlated better with objective measures and this was not observed for L1. Language skills that we learn formally in a structured manner have better reportability. Present study highlights the use of objective measures of language proficiency. The factor structure of L1 and L2 showed different patterns of clustering within and across domains. Task based and not domain based clustering was observed for L1. L2 language skills showed clustering within and across domains. L1 and L2 language proficiency also emerged as a continuous rather than a categorical variable. Importance of testing L1 language proficiency is also indicated. Conclusion: An objective measure of language proficiency in both L1 and L2 is essential for the selection of bilingual participants as the variations in the level of bilingualism may mediate systematic effects on bilingual language processing. There is a need to revise, how we define and characterize bilingual population for better clinical and research implication. Keywords: Bilingualism, language proficiency, tool.

50 51

52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92

Introduction Bi/multilingualism is a multifaceted phenomenon and thus requires greater understanding of bilingual population, bilingual language acquisition, measures of bilingualism, language representation etc. It is the understanding of bilingualism which influences our observations in experimental work on the interaction between language processing and other cognitive processes such as attention, memory and executive function. Bilingualism is influenced by social and environmental factors as well as the mode and pattern of language acquisition and learning. It is well known that India is one of the largest bi/multilingual countries. Indian constitution (8th Schedule) lists over 18 languages for official\administrative purpose, while more than 1652 mother tongues were reported to be spoken in India according to the Census of India 1961. Many of these languages and dialects may or may not have scripts but are in active use. The nature of bilingualism is also not the same across the country. To deal with complex linguistic situation in countries like India, where many people are not only bilinguals but multilinguals and there is prevalence of different language combinations, it is important to come up with some consensual decision about the pattern of acquisition of the languages known, measures of bilingualism and the phenomenon of bilingualism itself. It is not only the variability in the organisation of language skills in L2 but also in L1 that may influence bilingual language processing and may also interact with other cognitive processes. Third and important issue while dealing with bilingual population is with respect to the method of selection of bilingual participants and to measure the degree of bilingualism. Measuring bilingualism would also help in the prediction of performance on a particular language in experiments that involve bilingual language processing (lexical access and lexical selection, phonological processing, semantic judgement or syntactic processing). Finally, it would be interesting to see how L1 and L2 influence each other particularly in the context of Hindi English bilinguals where L1 is acquired informally and L2 is primarily learnt through formal instruction. In addition the order of acquisition of language skills across the domains of spoken/understanding and reading/writing is also different for both. For L1 (Hindi), spoken/understanding skills are acquired first and reading writing skills are acquired with formal schooling. On the other hand, L2 acquisition starts with literacy skills first with formal schooling and spoken/understanding follows and is mostly associated with literacy skills. Language use also varies for L1 vs L2 across contexts which were also measured through self report and it would be interesting to find if language use along with different acquisition patterns of spoken/understanding skills would predict the degree of bilingualism among bilingual participants. In a larger context, we could also re-examine the methodological traditions which are prevalent in Bi/Multilingualism research and how reliable and valid are these methods for classifying/profiling bilinguals. Looking at the complexity of bilingual language context in India, it is of utmost importance to have appropriate and effective screening tool for defining, classifying and categorizing bilinguals, i.e. by using self reported questionnaire, bilingual language tests tapping various language skills or use of individual language tasks. Such a tool would help screening bilinguals in clinical population as well as in bilingual research. Age of acquisition, language use as well as language proficiency are the measures which are used for such kind of categorization. Research on

93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134

second language (L2) acquisition has long been focusing on the impact of age of acquisition (AOA) on second language proficiency. Much evidence points to the notion that AOA is a primary predictor for the eventual proficiency level in L2 (Johnson & Newport, 1989; Stevens, 1999; Weber-Fox & Neville, 1996). At the same time, many studies regarding L2 acquisition underline the importance of other variables, such as extent of use of the language, and level of first language (L1) proficiency (Herman, Bongaerts, De Bot & Schrender, 1998; Flege, YeniKomshian, & Liu, 1999). Both age of acquisition and language use contribute in varying amounts to language proficiency and this proficiency of an individual in a language can be considered as an important measure of bilingualism. It is also important to understand that to what extent and in what way bilingual proficiency may coincide with variation in other variables, such as dominance, age of acquisition and use of language, which are equally important for the participant selection process. Because of the diversity in the selection criteria one can find diversity in results too, which led to initial finding of bilingualism being a consequence. Most of the underpinnings in bilingual research started with bilingualism as a consequence (Saer, 1923, Darcy, 1946, cited in Bialystok, 2001). Since last 30 years bilingual advantage over monolinguals in various cognitive abilities has been appreciated (Peal & Lambert, 1962 as cited in Bialystok, 1997; Ben- Zeev, 1977; Costa, Hernander & Sabastia-Galle, 2008; Siegal, Lozzi & Lurian, 2009; Bialystok & Feng, 2009). From a methodological point of view comparison groups have changed from time to time depending on the focus of the study. Many recent studies in the field of cognitive science compare monolinguals and bilinguals (Bialystok & Feng, 2009; Siegal, Iozzi & Lurian, 2009; Costa, Hernander & Sabastia-Galle, 2008). However, in linguistic context such as in India where bilingualism/multilingualism is a norm, it becomes very important to determine how we profile a bilingual as the comparisons will be made within a group of bilinguals. In an attempt to resolve the issue of experimental control, Hakuta and Suben (1985) in the past, highlighted the work by Ducan and De Avila (1979) suggesting that comparison of sub groups within the samples of bilinguals could act as a valid method of control in an experiment, which was further emphasized by Hakuta and Diaz (1985). Most of the work stated here has predominantly considered bilinguals from two extreme populations varying with respect to a particular aspect (age of acquisition: early vs. late, proficiency: balanced vs. unbalanced/ high vs. low proficient; or based on use: i.e. to consider them as dominant bilinguals). Clinical research also requires information about the level of bilingualism and language status to provide a sound base for assessment and rehabilitation of various communication problems. Such work related to participant selection is lacking in Indian context. It was further emphasized by Grosjean (1998) that due to lack of sufficient information regarding the bilingual participant selection procedures it is difficult to perform cross study comparisons as well as to replicate previous studies. Lack of sufficient details about the bilingual population may also make it difficult to explain the contradictory findings in literature for example the studies on effects of bilingualism on cognitive control. The effects observed in such studies are particularly influenced by the nature of bilingual population that was examined. Language proficiency has emerged as a useful and informative measure of bilingualism and it refers to the degree to which an individual exhibits control over the use of the rules of a language for one, some, or all of its numerous and diverse aspects. These include the phonological, syntactic,

135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175

lexical and semantic systems and discourse and stylistic rules for oral and written communication for different variations of a given language in various domains and social circumstances. In the past, there have been many studies which take into account different skills as a measure of language proficiency which include: confrontation naming, read aloud task (Oyama, 1976) grammaticality judgment task (Johnson & Newport, 1989), verbal fluency, category generation task, oral comprehension task (Bahrick, Hall, Goggin, Bahrick and Berger 1994) and self rated questionnaire (Coppieters, 1987; Marian, Blumenfeld, & Kaushanskaya, 2007). Grosjean (1998) suggested certain requirements as necessary while profiling bilinguals, which apart from biographical data include language history and language relationship, language stability, function/use of languages, language proficiency and language modes. The diversity in the testing methodologies, definition of bilingualism and measures of bilingualism may yield diverse results which may lead to lack of consensual opinion. To resolve such issues current paper hypothesizes language proficiency as a valid and reliable measure of bilingualism. Language proficiency has many dimensions to it, i.e. complexity at societal level as well as individual level and different methods were used in past to measure language proficiency (Marian et al 2007, Flege, Mackey and Piske 2002, Jia, Aaronson and Wu 2002). In a simplistic manner, language proficiency can be considered as the ability of an individual to speak or perform in a particular language. It is also important to consider different ways of defining and reporting language proficiency. So far it is tested using self reported questionnaire and a few objective tasks like Peabody Picture Vocabulary Test, translation test, grammatical judgment test etc. Most of the studies in this field predominantly focus on the L2 language proficiency and not much importance has been given to L1 per se. Most of the research with respect to experimental work on cognitive aspects of bilingualism in India has depended upon self reported proficiency with occasional use of an objective task. However, the assessment tools that have been employed are based on the literature in the west and not based on any objective data on the nature of bilingualism in Indian context. Hence, we examined both L1 and L2 proficiency as well as to lay out the factor structure of L1 and L2 among Hindi-English bilingual adults. This kind of ground work has not been done in Indian context. In addition we hypothesized that language proficiency would emerge as a continuous variable. Many bilingualism researchers use questionnaires to assess L2 language proficiency along with an objective measure. There is, however, little to know what these questionnaires measure, how precise they are, and how they relate to actual language proficiency more objectively. This is what we examine in this paper by investigating the correspondence between self report and a comprehensive objective assessment of language proficiency. We also examine the factor structure of L1 and L2 across language skills. In addition, we compare L2 performance to similar tests in L1, to see to what extent both are the same. Based on the objective and subjective performance of bilinguals, we aimed to analyze the L1 and L2 interaction within and across language skills among H-E bilingual adults. Subjective test could provide us with biased self reported information; justifying subjective information by some form of objective measure could certainly strengthen the way bilingualism is considered in

176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212

a particular study. Secondly we also examined variability in L1 and L2 proficiency looking at the factor structure of both L1 and L2 within and across language domains. Thirdly, the usability of the tool in participant selection for experiments involving bilingual population was also examined. We hypothesize that using both subjective and objective information may provide wider information and few objective tasks may be more appropriate than the others. Finally, such a tool may also aid in better prediction of behavioural and experimental performance and in understanding the activation and suppression mechanisms in bilingual language processing as a function of basic organization of language skills in bilinguals which is not the primary focus of the present paper. It would also enable us to interpret the task specific or nonspecific effects when the stimuli are processed at phonological/lexical/semantic/syntactic levels. Methodology This study aimed to explore the nature of language proficiency in L1 (Hindi) and L2 (English) through the use of modified version of Language Background Questionnaire and indigenously developed test of language proficiency. Organization of language skills across domains for Hindi and English was also examined. Participants The participants for the present study were 85 Hindi-English Bilinguals (M=20.84 years, range of 18 years to 26 years; 52 males and 33 females). Only those participants were selected who had completed a preliminary screening which indicated that their L1 was Hindi and L2 was English, which was used on day to day basis and they had at least 7 years of basic education in both the languages with no significant history of sensory, motor or neurological disorders. Participants were selected randomly from Allahabad, India. They were interviewed regarding their language use and exposure and were administered the test of language proficiency. Material In order to meet the purpose of this study, the following instruments were used: Language Background Questionnaire was employed to get information about the languages in use, frequency of use; self reported proficiency, linguistic environment at home, work etc. Domains assessed in the present questionnaire include acquisition history (age of acquisition and at what age they became fluent), contexts of acquisition (modality=oral/written/both and environment of acquisition=informal/formal/both), present language use (in percentage), language preference (1-3 rating scale; where 1= never, 2= sometimes, 3=most of the time) and proficiency rating (0-10 rating scale). Apart from these questions contribution of various other factors such as use of language with family, friends, extended family, neighbours will be assessed by asking participants to name the language predominantly used along with hours of usage (per day). Participants will also be asked to indicate the medium of instruction and self reported proficiency level on different domains (1-5 point rating). The administration of questionnaire takes approximately 30 minutes.

213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250

An indigenous Test of Language proficiency in Hindi and English was also employed to examine language proficiency in Hindi and English. Proficiency in all domain of language function (i.e. speaking, understanding, reading and writing) was examined. The test of language proficiency consists of two sections under the following domain: Speaking/ Understanding domain and Reading/ Writing domain. Each of the domains is explained along with subskills. Speaking/ Understanding domain The goal of the tasks under this domain is to tap the crucial aspects of bilingual language proficiency, in terms of speaking and understanding skills which indicates oral and aural proficiency. Hindi and English versions were matched with respect to kind of task, number of items as well as scoring method. Appendix 1 would elaborate the scoring process in the supplementary material. Confrontation Naming Task: In this task participants were provided with 30 pictures consisting of high as well as low frequency nouns. And the participants were asked to name them when shown. Pictures used for confrontation naming were taken from IPNP (pictures by Abbate & LaChappelle (1984)), developed by the UCSD. Spoken discourse task: In this task participants were instructed to describe a picture carefully by focusing on the overall theme of the picture along with individual items in that particular picture. A grand rubric score (appendix 1a supplementary material) is calculated by summing the scores on the following aspects: overall impact and achievement of purpose (whether the participant establishes main idea), organization and techniques (coherence and cohesion with test, method of organization) and mechanics (focusing on grammar, pronunciation, presence of pause). Pictures were selected from Boston Diagnostic Aphasia Examination (Goodglass & Kaplan, 1983) and Western Aphasia Battery (Andrew Kertesz, 1987) for English and Hindi respectively. Auditory comprehension task: This task was chosen to assess participants ability to understand the content aurally. To assess, 5 questions were asked related to the passage and the questions were formulated based on main idea of the passage, cause and effect relationship and inference generated. Passages were selected from Test of language proficiency: Hindi (Pon Subbiah, 2005) and web description respectively for Hindi and English version. Convergent production task/synonym task: This was the only section which had 2 different tasks for both Hindi and English version with similar scoring process. In the English version convergent production task was administered which requires the participants to name as many meaning possible of a particular word (Thorum, 1986). E.g. By*/ bye/buy can have three different meaning: to purchase, preposition- by and Good bye. Hindi counterpart for this section was synonym task where the participants were asked to name as many synonyms (at least 3) of a particular word (Pon Subbiah, 2005).

251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287

Under this subtask there is difference between languages but it intends to tap participants ability to demonstrate their knowledge of word meaning. Reading/ Writing domain Reading comprehension task: This task attempts to assess the ability of the participant to read and understand what is stated or implied in a written passage, and then to answer questions based on it. Each passage contains 159 and 208 words respectively in Hindi and English part of the test (Shipley & McAfee, 2008; Pon Subbiah, 2005) which was followed by 5 questions. The questions were formulated to tap main idea of the passage, cause and effect relationship and inference generated in the passage. Reading fluency measure: The passage read in the previous task was used to assess reading fluency by using Fluency Rubric (Zutell & Rasinski, 1991). A grand rubric score is calculated by summing the scores on the following aspects: smoothness and pacing (with respect to punctuations and break), confidence, accuracy (with respect to pronunciation) and expression (with respect to change in voice with different content). (Appendix 1b supplementary material) Phonological awareness skills: In this section participants were asked to perform simple meta-linguistic task at sound and syllable levels. The tasks involve segmentation, blending, rhyming and counting of sounds. It was kept in consideration to keep the subtask between languages. Written Discourse analysis: In this task participants would be asked to write a procedure asked for. The participants were clearly instructed to write in paragraph and have a good beginning and ending lines. A grand rubric score (appendix 1c) is calculated by summing the scores on the following aspects: overall impact and achievement of purpose (whether the participant establishes main idea), organization and techniques (coherence and cohesion with test, method of organization) and mechanics (focusing on grammar, spelling mistake).

Administration of the proficiency test in both languages takes around 45 minutes. Scoring method along with the task are attached in appendix 1 supplementary material. Administration of the language background questionnaire and test of language proficiency was completed in a quiet testing room. Order of administration was counterbalanced across participants with respect to Hindi and English languages. Recording for the discourse sample and reading sample also occurred in the testing room. Such kind of measure was used to employ some amount of objectivity to the process of assessment and it was not intended to define proficiency and group the bilingual individuals. Total scores were calculated within each section under the domains of speaking and understanding scores which is a combination of spoken discourse score, confrontation naming score, convergent production and auditory language comprehension score. Similarly total reading and writing score consisted of reading

288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326

comprehension, reading fluency score, phonological awareness score and written discourse score.

Results The purpose of the present study was to explore the relative contribution of language proficiency in profiling bilinguals in a continuous manner (rather than categorical) in Indian context and to determine important measures of language proficiency. A modified version of Language Background Questionnaire (Vasanta, Suvarna, Sireesha, & Bapi Raju, 2010) and indigenously developed Test of language proficiency was administered on 85 H-E bilinguals. Normal probability distributions were derived to find out, whether language proficiency behaves as a continuous variable. Factor analysis was performed to establish the factorial validity for the language proficiency test as well as to find out how different tasks were correlated to each other within and across languages. Correlation and regression analysis were performed within and across languages as well as between subjective (self report) and objective measures. Cronbach s alpha was calculated as a measure of reliability for the language proficiency test. Language Background in L1 and L2 Information based on language background questionnaire (i.e. subjective information) was categorized as: a) age related information which consisted of age of acquisition in school and age of fluency in reading and writing skills, percentage of exposure; b) language use related information included task based scores; c) proficiency related information comprised of self reported measure of reading, writing, speaking and understanding skills. For language background information mean scores and standard deviations were computed which are presented in Appendix: 2. Data based on language background information suggested that exposure to Hindi was since birth at home setting, whereas English was predominantly introduced in the school years. As expected, order of acquisition followed the trend of native language being the first acquired language. Whereas, order of dominance was suggestive of Hindi being the dominant language except for 4 participants. Age related information: It was interesting to know that there was a lot of variance within the group with respect to age of learning L2 in school, which is suggestive of different timeline for learning L2 in school ranging from 4 to 11 years (Mean (L1) = 4.32 years; Mean (L2) =5.37 years). And this variance increases in both L1 and L2 while reporting the age at which participants became fluent in a particular language (Mean (L1) = 7.37 years; Mean (L2) = 10.786 years). Subjective judgment of age as a measure leads to variation because of the bias in participants recall of the information, movement of some participants from villages (where Hindi was the only language taught) to cities. There was no significant difference between questions targeting percentage use in contrast to percentage exposure. Participants reported their language use on a particular task on a 3 point rating scale. Participants reported predominant use of L1 in their day to day life (M (L1)= 64.49% M (L2)= 35.44%), where as language use with

10

327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365

respect to day to day tasks suggest of similar mean scores (M (L1)= 2.29, M (L2)= 2.06). There was no significant difference between L1 and L2 on reading (M (L1) = 4.3, M (L2) = 4.1) and writing (M (L1) = 3.8, M (L2) = 3.9) self reported proficiency scores. Whereas speaking (M (L1) = 4.5, M (L2) = 3.3) and understanding (M (L1) = 4.7, M (L2) = 4.0) self reported proficiency scores showed a difference between L1 and L2. Relationship between age of acquisition, language use, self reported proficiency and objective measure of language proficiency Bivariate correlation and regression analysis was performed to evaluate the degree of relationship within and across objective (i.e. task performance) and subjective measures (self reported measure) of language proficiency. The purpose was to find out whether self reported measure of age related information, use of language as well as self reported proficiency could predict performance on objective tasks of language proficiency in the domain of speaking/understanding and reading/writing in both L1 and L2. Results of correlation and regression analysis are discussed with respect to domains i.e. speaking/understanding followed by reading/writing domain. Multiple regression analysis was performed by using simultaneous method as there was no theoretical model. On initial inspection the total score on speaking/understanding domain for L1 showed statistically significant correlation, r = .25, p < 0.05, n = 85, with only one predictor variable under the broad subgroup related to self reported proficiency (composite score of self reported performance on certain language tasks in L1). Self reported measure of proficiency in the form of composite scores on tasks predicted 6.7 % of variance (R square =.067, adjusted R square =.055). The model was significant, F (1, 84) = 5.92, p < 0.05. For L1, within the speaking/ understanding domain, age related information showed no correlation with the task performance while use related predictor variables and spoken discourse task indicated a low but statistically significant negative correlation, r = -0.26 and -0.31, p < 0.05 and p < 0.01 respectively, n = 85. Since, predictor variables such as self reported reading, writing, speaking and understanding proficiency did not correlate with speaking/understanding tasks, they were not subjected to regression analysis, as shown in table 3a. Insert table 3a about here Insert table 3b about here By using simultaneous method of regression analysis, model including use related predictor variable and percentage of exposure in L1 predicted 12% of the variances on spoken discourse score (R square = 0.12, adjusted R square = 0.129) and was significant, F(2, 82) = 6.08, p < .01. As shown in table 3b, self reported reading and writing proficiency in L1 and self reported composite score on proficiency tasks had statistically significant but weak correlation with synonym production task, r = .240, r = .282 and r = .395, p < 0.05, p < 0.01 and p < 0.01 respectively, n = 85. Multiple regression with self reported reading and writing proficiency and self reported composite score of the task description, predicted 17.4% of variance (R square =

11

366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402

0.174, adjusted R square = 0.144) for synonym production task as a criterion variable. The model was significant, F (3, 81) = 5.704, p < 0.01. In the domain of reading and writing for L1 there was no significant correlation between total scores on the objective tasks and the subjective measure of proficiency. Reading fluency in L1 was significantly correlated with self reported measure of reading and writing proficiency with r = 0.253 and 0.327 at p < 0.05 and 0.01 respectively for n = 85. Further, self reported measure of reading and writing skills predicted 10.8% of the variance, (R square = 0.108, adjusted R square = 0.086) and was also significant F (2, 82) = 4.94, p < 0.01. Insert table 4a about here Insert table 4b about here Results presented in table 4a indicate that use related information (percentage of exposure and composite score of the task based scores) predicted 16.4% of the variance in the total scores of L2 speaking understanding domain (R square= 0.164, adjusted R square= 0.144) and the model was significant, F(2,82)= 8.050, p < 0.01. On the other hand, proficiency related self reported information could predict 32.5% of variance in total score of speaking understanding task performance (R square =0.325, Adjusted R square= .282) and the model was significant with F(5,79) = 7.599, p < .001. Similar analysis was done for reading writing domain and age of acquisition information were excluded from regression analysis because of lack of correlation as shown in table 4b. Use related self reported information predicted 33.3% of variance in total score in reading/writing domain (R=0.333, R square=0.111, adjusted R square=.089), whereas self reported proficiency predicted 31.8% of variance in the total score (R =.564, R square= .318, adjusted R square= .275) and both models were statistically significant. Insert table 5a about here Insert table 5a about here Finally, correlation analysis was performed between task scores of L1 and L2 for speaking/understanding as well as reading/writing domains. Spoken Discourse analysis for L1 correlated with most of the L2 speaking understanding tasks. There was low but statistically significant correlation between speaking understanding total score for L1 and L2 (Table 5a, 5b). To sum up, age related information as a predictor variable showed no correlation with any of the domains or languages. Use related information and proficiency related information (i.e. percentage exposure, composite task based scores for use and self reported proficiency on speaking, understanding, reading and writing) predicted performance on respective languages. Use related information and proficiency related information on self reported questionnaire as a combined predictor variable were able to account for maximum variance on L2 tasks as compared to L1; similarly scores in reading/writing domain were able to account for the variance more appropriately as compared to speaking/understanding domain.

12

403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 Insert table 1a about here Insert table 1b about here Factor analysis across sub skills in L1 and L2 Factor analysis was performed to learn if the observed variables can be explained largely or entirely in terms of factors which are a cluster of 2 or more individual variables. It informed about the nature of domains with respect to sub skills and also whether the tasks across and within domains as well as across and within the two languages could be combined. Within domain (i.e. speaking and understanding domain or reading and writing domain) Under spoken-understanding domain for L1, 6 components were extracted from the data set by means of factor analysis (Table 1a). Of these factors, the first 2 components had Eigen values greater than 1 and accounted for 62.6 % of all variance. Both the components account for 33.04% and 29.6% of variance respectively. Under the domain of speaking and understanding in L1, tasks on oral and aural aspects of discourse were clustered under one factor and tasks on naming skills (semantic aspects of language) clustered as the other. All the tasks under spokenunderstanding domain in L2 emerged as one component. Similar findings were also evident for reading and writing domain, where L2 had one single component whereas factor analysis of L1 reading and writing tasks led to 3 components. Out of 6 initial components that were extracted, first 3 had Eigen values greater than 1 and accounted for 79.71% of variance. Table 1a presents all these components. 30.96%, 26.9% and 21.76% were the respective variances of the 3 components for L1 reading and writing tasks of proficiency. Discourse tasks in reading writing domain also clustered together under one component. As phonological awareness task intends to measure the metalinguistic ability of the individual, its presence as a separate component was probable. Within domain factor analysis of L1 showed different components. Discourse skills (production and comprehension) and naming skills (convergent production, confrontation naming) emerged as two components in speaking/understanding domain. Discourse (reading and written discourse understanding), reading fluency, speed and phonological awareness were separate components in the reading/writing domain. L2 tasks clustered under one component for both speaking/ understanding and reading/writing domains whereas L1 tasks showed clear distinction between different language skills i.e. discourse skills, semantic skills (naming, convergent production), reading skills (fluency and speed) and metalinguistic skills (phonological awareness).

13

438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473

Across domain (i.e. speaking and understanding domain vs. reading and writing domain) 12 components were extracted from L1 data set by means of factor analysis. Of these factors, the first 5 components had Eigen values greater than 1 and accounted for 74.71% of all variance. These 5 components were assigned construct names indicative of characteristics and are listed in terms of variance in Table 2a. There was no segregation with respect to domains rather it was task specific for L1. Naming tasks were grouped under one component, whereas L1 discourse tasks were clustered according to whether it was a production or an understanding task. Similarly, L1 showed more scatter as compared to L2.

Insert table 2a about here Insert table 2a about here

Tasks in L2 yielded 3 components. First component accounted for maximum of 43.41% variance and included all the tasks of proficiency for L2 except for phonological awareness, which was the only variable in 2nd component explaining 14.88% of variance followed by the 3rd component comprising of discourse understanding in both domains (Table 2b). For both the languages across domains distinction was not domain specific whereas it was specific to subskills/tasks i.e. production tasks as one and comprehension tasks as other. Influence of L1 and L2 language skills on each other In an attempt to find out the influence of both L1 and L2 on each other on the test of language proficiency, regression analysis was conducted. On speaking understanding domain total score of L2 was able to predict 21.2% of variance with respect to total score of L1(r= 0.461) and the model was significant with F = 4.202, p=0.002. Same was not observed from L1 to L2 i.e. L1 total scores were not able to predict performance on L2 language proficiency tasks. However, performance on L1 reading writing domain predicted performance on L2 and vice versa. Total scores of L1 and L2 both showed high correlation with task of reading and writing r= 0.512, r = 0.501 respectively. Results based on Normal Probability Distributions with Language Proficiency data Language proficiency tests scores across the spoken, understanding, reading and writing domains were subjected to statistical analysis to achieve the objectives mentioned above. Over the years, general practice in bilingual research has been to group bilinguals into high and low proficient bilinguals, or early and late bilinguals etc, basically to compare two extreme ends of the same population. But to do so, the bilingual population when picked randomly should show a bimodal distribution. Based on our results, it was interesting to know that the participants fall in a continuum rather than categorically in a distribution on the basis of tasks on speaking/understanding and reading/writing domains in L1 and L2.

14

474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511

In an attempt to address the issue of language proficiency being continuous or discrete variable Q-Q plots as well as normal probability distributions were plotted for the total scores in speaking/understanding domain and reading/writing domain in both L1 and L2. The Q-Q plot shows the expected distribution of the standardized observed value (X axis) across the total scores which were compared to the expected normal values (Y axis). Any deviation from the X=Y line implies deviation from the normal distribution, which is not the case as evident in the figure 1 and 2.

Insert Figure 1 about here

To sum up, results indicate that total scores of L1 as well as L2 on the measures of speaking/understanding domain and reading/writing domain showed normal probability distribution. Validity and reliability of the test of language proficiency Factorial validity and reliability (by measure of Cronbach s alpha) were also measured to assess the extent to which tasks captured similar information (i.e. language proficiency). Factor analysis showed that grouping of each task under different component makes intuitive sense i.e. L1 tasks clustered into meaningful groups, which was evident through the grouping of discourse tasks as one common component, task requiring semantic information access were grouped as another, whereas L2 tasks formed single construct for both speaking/understanding as well as reading/writing domain. Cronbach s alpha as a measure internal consistency was found to be 0.646 for 24 items including both L1 and L2 tasks together, while considering L1 and L2 separately it was 0.430 and 0.658 for L1 and L2 respectively. Discussion With the expanding scope of bilingualism, cognitive and linguistic aspects related to bilingualism have gained prominence gradually. In this context, it is increasingly important to understand and validate the methods for participant selection in bilingual research. Bilingualism research in the felid of cognitive science and psycholinguistics focuses on mechanisms related to attention, inhibition and cognitive control in bilingualism (Kroll, Bob, Misra and Guo, 2008; Colzato, Bajo, van den Wildenberg, Paolieri, Nieuwenhuis, La Heij, & Hommel 2008; Chauncey, Holcomb and Grainger, 2009; Bialystok, 2007) second language acquisition (Van Hell & Tokowicz, 2010), bilingual lexical representation, (Kroll & de Groot 2005), models of language processing (Dijkstra & Van Heuven 2002) and bilingual memory representation (Kroll & Stewart,1994). A lot of clinical research focuses on the role of bilingualism on assessment and management of language disorders, especially in countries like India. For instance, clinical research on Indian population emphasizes upon the role of proficiency in language recovery in Aphasia and there is also a greater support towards use of bilingual mode for the rehabilitation (Chengappa, 2009). Three of the most discussed variables in bilingual research are namely age of

15

512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554

acquisition, use of language and language proficiency that may influence the experimental data. These are mostly used while profiling bilinguals or while categorizing bilinguals. Age of acquisition can be quantified by the participants with more assurance. However, use and proficiency are acquired overtime and hence are likely to be inconsistent in reportability. Language proficiency may be considered as a functional outcome of the interaction between age at which language is acquired or learnt and percentage of exposure to that particular language. In spite of age related information being discrete in nature, as in the current study there was lack of correlation between age of acquisition (AOA) with performance on language skills which raises a concern about the use of age related information as a holistic measure of bilingualism. Language proficiency as a measure can be a viable option while profiling bilinguals as it is influenced by language use and age related information to varying degrees. Language use on other hand is influenced by language experience and language mode which makes language use a very dynamic variable. Relying on language use would require us to satisfy the experiential requirement while assessing it. Advantage of language proficiency is that it can be quantified, which has been done in the past except for the fact that only certain skills (like confrontation naming, comprehension, translation etc) were considered. Current test of language proficiency not only examined language skills at semantic, syntactic and discourse level in the domain of both speaking/understanding and reading/writing but also did so in both the languages (L1 & L2). Test of language proficiency in Hindi and Englsih can be considered as a composite measure of language proficiency along with other self reported measure. As it was evident in the current study that the expected level of proficiency varied across the language skills in native language also, that may be because of the varying patterns of native language use. The idea of testing both the languages comes from the fact that in today s bi/multilingual world effective communication demands for a trade off between the languages which influences the language use pattern, and this is subject to individual variations. One would expect different levels of bilingualism based on the trade off between domains as well as between languages. For example, one person would be categorized as high proficient in spoken/understanding domain in both his languages and low proficient in reading/writing domain in his second language. On the other hand, we may also have a bilingual who is at the intermediate level of proficiency in native language and highly proficient in second language. This sort of distinction would create a 6*2 matrix with individual language related level (i.e. high, mid, low) vs. two domains. This has been theoretically described in Bialystok and Cummins work, where Cummin (1991) accounts for two different types of language proficiency which includes Basic Interpersonal Communication Skills and Cognitive Academic Language Proficiency. In the current study, we focus on continuous nature of language proficiency rather than categorizing language proficiency in different levels. Bilingualism can be characterized as continuous variable because the factors associated with bilingualism i.e. language proficiency also functions in a continuous manner and thus should be treated as a continuous variable. Such an approach may help us understand the pattern of laboratory based experimental results. There may be task specific variation in a particular experiment (naming words, reading, lexical decision) due to changes in levels of proficiency in both the languages. In this context, we discuss the findings of the present study based on our data with the help of Test of Language Proficiency and Language History Questionnaire.

16

555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595

Correlation between subjective and/or objective measure Many of the psycholinguistic/sociolinguistic studies define bilinguals only on the basis of questionnaire based information (Coppieters, 1987 and Marian et al, 2007). One such attempt by Marian et al (2007) in the form of an assessment tool resulted in the development of LEAP Q (Language Experience and Proficiency Questionnaire), where they have probed separately for language proficiency, language dominance and language preference. This questionnaire elicits proficiency ratings in speaking, listening, reading and writing domains. Reliability of the self reported measures is a concern. In the current study, it was evident that age related information had no significant correlation with objective performance on Language proficiency tasks whereas use related information and proficiency related information could account for less than 50% of the variance on the objective task of language proficiency. There was a correlation between speaking/understanding domain of L2 and age of fluency in reading and writing which can be attributed to the fact that fluency is achieved at a later age and therefore the reportability improves for the same. Apart from the fact that L2 emerges as a robust variable in itself, i.e., most of the L2 task constituted a common factor as well as overall scores of L2 were correlated in different domains. It is also interesting to note that reportability of language that is acquired is less than the one that is being learnt in a structured setup. Complete reliance on the subjective measure may not provide us with adequate information about bilingualism rather inclusion of objective tasks along with the questionnaire will enhance the profiling of bilinguals. Moreover, both objective and subjective measures are conceptually related to each other, which help in understanding the relationship between variables like language use, exposure and specific language skills. The overall day to day use of a language weighs more than age related information and that it is more for the L2 as compared to L1. In the current study, out of all the tasks of language proficiency, confrontation naming, discourse analysis and reading comprehension were found to be better correlated with subjective measures and thus can be used as screening measures of language proficiency in case of a quick assessment in experimental setup. Though we would advocate the use of the entire tool for measuring language proficiency as L1 showed task based clustering and L2 showed task/domain independent clustering. It is interesting to highlight that self report of L2 corresponds well with objective performance of L2 which is not the case with L1. Age related factors do not predict performance on language skills across domains for L1, same was predictive in case of L2 supporting the previous claim of stronger reportability of age related factors in the language which is learnt in a structured setup. Language proficiency: as a measure of language skills There are predominantly three aspects to bilingualism: age of acquisition, language use and language proficiency. In the current study, language proficiency stands out as an appropriate measure of bilingualism. It is able to objectively tap different language skills in both L1 and L2. There are many points of view on language , which include the formal linguistic view, cognitive-functionality view and socio-cultural view (Goto Butler & Hakuta 2004). This paper considers language from cognitive-functionality point of view which considers language acquisition to be data-driven and as a part of general cognition. Therefore, the tasks examine

17

596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636

different aspects of language skills, not just at the level of words or sentences but also at a higher level such as discourse. A lot of work considers language proficiency as an independent variable in experimental studies related to bilingualism. Following are the skills which have been assessed in the past to define language proficiency, comprehension and production of oral and written language, translation task, read aloud tasks (Oyama, 1976), grammaticality judgement task (Johnson and Newport, 1989 & Fledge, Mackey and Piske, 2002), verbal fluency (Bialystok, 2007) and self reported questionnaire (Coppieters, 1987 and Marian, 2008). The current paper presents a tool that taps all these skills in the form of one comprehensive tool. In the current study, total scores of language proficiency correlates in a better manner as compared to individual tasks. It further enhances our claim of using a composite score of different tasks that assess various aspects of language. We developed a set of language proficiency measures in both the languages i.e. Hindi and English, keeping them equivalent both in terms of validity and reliability while considering language proficiency as an important variable in defining bilinguals in accordance with Butler and Hakuta s guidelines. Results also suggest that performance on the test of language proficiency show a lot of variation in both the languages within and across domains. It was interesting to know that L2 tasks in the domain of speaking/understanding as well as reading/writing were clustered under one common factor, which was not the case for L1. Such kind of clustering is observed as L1 is acquired and L2 is learnt through instruction. Language which is acquired has no strict way of experiential pattern, and the dynamic nature of this exposure leads to the variability in reportability of acquisition related information whereas, L2 is learnt through instruction and hierarchical increments in every grade adds some sort of consistency while learning the second language. Such kind of clustering indicates more overlap within the tasks for L2. Factor analysis with both L1 and L2 tasks put together, showed task specific segregation of performance rather than domain specific (i.e. all production task clustered together instead of speaking/understanding tasks getting together). This also shows that within a domain there is enough of variation to assess different aspects of language performance. This implies that language organisation is task specific (i.e. naming task, comprehension task etc) and reading/writing and speaking/understanding as domains cannot be segregated. It indicates that our language representation is interwoven in both the domains i.e. speaking/understanding and reading/writing. Oral and aural segregation in tasks may indicate difference at the level of the underlying mechanisms. Experimental work with bilingualism may get influenced by such findings in a way that researchers with certain research questions may expect a task specific (visual word recognition, spoken production) correlation with a measure of language proficiency. Our preliminary findings with experimental data suggest a task specific correlation with the measure of inhibitory control. In L1, reading comprehension predicted inhibitory control performance while total score on L2 task predicted performance. It is interesting to note that on factor analysis we find a lot of scatter in the L1 tasks which may implicate the use of a comprehensive tool including all language domains for the determination of language proficiency in L1. Such kind of profiling will be different for L1 and L2 because of the way the subtasks are clustered together for a particular language.

18

637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678

In addition, correlation between L1 and L2 on the discourse task suggests that knowing L2 enhances the skills at discourse level for L1, which is nothing but organization of information in L1. It s been implicated in many studies that L1 acquisition helps in L2 performance indicating a cross language transfer (Sparks, Patton, Ganschow and Humbach, 2009). However, in the current study there was an overall correlation of discourse task in L1 with all L2 tasks. Although L1 discourse did not predict L2 discourse performance but L2 discourse predicted L1. Thus the ability to process two languages appears to enhance the overall higher level language skills such as discourse production. This sort of correlation in L2 to L1 tasks at discourse level is suggestive of enhancement of language organisation skills through the learning of two languages. The fact that L2 showed higher correlation within and across domains as compared to L1 is supported by the finding that L2 language skills clustered as one common factor as compared to L1 language skills. Language proficiency: categorical or continuous variable The very nature of treating bilingualism as a categorical variable by manipulating language proficiency, language use and/or age of acquisition creates some sort of selection bias. Common concerns to the researchers are what kind of screening tasks should be prepared, selection criteria etc. Once the selection has been done on the basis of proficiency, age of acquisition or use of language, comparison between categories or levels of bilingualism takes place. The very nature of this sort of comparison (i.e. high vs. low, balanced vs. unbalanced etc) is biased because doing so actually means comparing the extremes which ought to show a difference. Such a comparison is viable if assessment of all bilinguals on certain aspects (AoA, proficiency and use) leads to a bimodal distribution. In contrast, our results suggest that language proficiency of a bilingual shows a normal probability distribution across domains or language skills, negating the traditional method of categorizing as high vs. low proficiency bilinguals on the basis of language proficiency. By doing so, researchers can assess a bilingual on desired aspects of language proficiency and instead of grouping them, thus treat proficiency as a continuous variable. Grosjean (1998) while enumerating many methodological and conceptual considerations in bilingual research emphasized on the demerits of considering bilingualism as categorical variable. Grosjean placed a lot of emphasis on consideration of bilingual assessment measure as a covariate during analysis. Bialystok (2001) also highlighted the continuous nature of bilingualism and considered it as an inherent property. Bialystok further emphasizes the need to consider both, the type of language proficiency (with respect to the domain) and degree of language proficiency in each language of a given individual during language testing. However, this has not been empirically examined. Any method of dealing with bilingualism should have the capability of ideally setting the boundaries of language proficiency as well as acknowledge variability, thereby provide some sort of metric position of the learner in certain skills. While suggesting a framework, Bialystok (2001) states the relationship among use of language and underlying cognitive requirements, where analysis of representation, structure and control of attention creates the orthogonal axes. Language tasks or language use fall on this Cartesian space indicating the dependence on each of these cognitive processes. Thus, grouping individuals based on some arbitrary cut off point, would automatically exclude certain individuals from the bilingual category though they have certain amount of skills in both their languages (Bialystok,

19

679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719

2001; Grosjean, 1998). By doing so, we have extreme ends of the population and it is easier to find differences between the group of bilinguals (i.e. high vs. low proficient) on any of the linguistic and cognitive tasks. As language proficiency varies on a continuum, it could better be treated as a predictor variable and need not be manipulated as a categorical/discrete independent variable. Another reason for considering language proficiency as a continuous variable is to remove the methodological bias which can be introduced by dichotomy. Future work needs to empirically test the validity of treating language proficiency as a continuous variable. Although our preliminary results based on our experiments on selection and inhibition show that total scores on language proficiency does predict performance on measures of inhibitory control. We also find that L1 proficiency shows a task specific prediction for the performance on experimental tasks, whereas L2 proficiency shows a domain based prediction of performance (Dash & Kar, 2011). In the above mentioned studies bilinguals were treated as one group varying in proficiency. In one of our previous studies we also had categorized bilinguals as high and low proficient bilinguals and had subjected them to the attention network task which combines the cuing and the flanker paradigms (Kar, Khare, & Dash, 2011). We found that though high proficient bilinguals performed better with respect to conflict resolution, however, low proficient bilinguals were slow on both congruent and incongruent trials. These findings suggested general decrements in performance (slower RTs) and hence the difference between the two groups could not be interpreted as benefits in executive control in high proficient bilinguals as compared to the low proficient bilinguals. Such preliminary findings need to be tested further in our future studies. Bilingual research has taken both the approaches i.e., categorizing bilinguals or treating levels of bilingualism as continuous variable, with greater inclination for grouping individuals into extreme categories particularly in the context of research looking at the benefits of bilingualism and language control and cognitive control. Considering language proficiency as a continuous variable will help us in considering the whole population rather than the extreme ends of the population in the form of high and low proficient bilinguals. It will also help to explain the interaction between bilingualism and cognitive processes such as attention, memory, and executive control. Test of language proficiency in Hindi and English: As an objective tool for use Lack of usage of a consistent tool for profiling bilingualism is evident in literature which may be a resultant factor for the variability seen in literature pertaining to bilingualism. The current paper is an attempt to conceptualize a tool development process by addressing the issues of reliability and validity of an objective measure of language proficiency applicable for bilingual research. Our results support the importance of such tool with good factorial validity and adequate reliability which can be used with little modification with respect to the language being examined. Use of such tool has both research related and clinical implications. The fact that bilinguals vary from each other in terms of language proficiency in a continuous manner, it demands for a radical step in re-evaluating information on bilingualism at conceptual and methodological level. Current study highlights the importance of testing both the languages, as the described continuity in the data is not just for second language rather exists in the native

20

720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754

language also. Native language also demonstrated the variability observed in L2 with respect to language proficiency, indicating the fact that language proficiency as a continuous variable in L1 may also have some sort of influence on our experimental results based on issues related to cross language transfer, lexical access, and inhibitory control and so on. However, the current study examined language proficiency in L1 (Hindi) and L2 (English) which is a combination where L1 has been acquired informally and is the dominant language in spoken understanding domain whereas L2 has been learnt formally through instruction for the population under study. Further, studies are needed on those combinations of languages where both L1 and L2 have been acquired informally. Findings of the current study need to be considered in theorizing about bilingualism as well as in experimental context but this may have applications in clinical domain as well. Clinically, matching of bilingual individuals with acquired communication disorder on the basis of performance on such a tool may act as an important attribute in clinical research. Clinical decision making while selecting participants for trials can be based on level of proficiency as it has functional importance rather than depending solely on neuro-anatomical findings. Idea is not to replace the previously used tool rather provide a composite tool. It emphasizes the use of such tasks addressing all aspects of language and treat language proficiency as a holistic variable. Such a tool may help to form guidelines for counselling for language selection in therapy for individuals with speech and language impairments.

Conclusion Language proficiency is one of the most important ways of characterizing bilingualism particularly with respect to the levels of bilingualism. It is an important correlate of the effects of bilingualism on language processing and other cognitive processes. Continuous nature of language proficiency in bilingualism demands for treating it as a predictor variable. It is evident that in current bilingual research, there is need for a comprehensive objective method of bilingual language proficiency assessment, not just in language that is learnt (L2) but also their dominant first language (L1). Most of the research focuses on L2 proficiency but interestingly enough, L1 proficiency varies in the same way as L2, this comprehensive method of testing is also required because of the fact that L1 and L2 have different factor structures. Finally this study highlights the use of test of language proficiency in Hindi and English as a reliable and valid tool.

21

755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772

References Abbate, M. S., & La Chappelle, N. B. (1984). Pictures, please! A language

supplement. Tucson, AZ: Communication Skill Builders. Bahrick, H. P., Hall, L. K., Goggin, J. P., Bahrick, L. E., & Berger, S. A. (1994). Fifty years of language maintenance and language dominance in bilingual Hispanic immigrants. Journal of Experimental Psychology, 123, 264:283. Ben-Zeev, S. (1977). The influence of bilingualism on cognitive strategy and cognitive development. Child Development, 48, 1009 1018. Bialystok, E. & Feng, X. (2009). Language proficiency and executive control in proactive interference: Evidence from monolingual and bilingual children and adults. Brain and Language, 109, 93 100. Bialystok, E. (1997). Consequence of bilingualism for cognitive development. In de Groot A. M. B. and Kroll, J. F. Tutorials in Bilingualism: Psycholinguistic Perspectives, 279-300. Bialystok, E. (2001). Bilingualism in development: Language, literacy, and cognition. New York: Cambridge University Press. Bialystok, E. (2007). Cognitive effects of bilingualism: How linguistic experience leads to cognitive change. International Journal of Bilingual Education and Bilingualism, 10, 210-223.

22

773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790

Chauncey, K., Holcomb, P. J. & Grainger, J. (2009). Primed picture naming within and across languages: An ERP investigation. Journal of Cognitive, Behavioral, and Affective Neuroscience, 9, 286-303. Chengappa, S. (2009. Bi/Multilinualism and issues in management of communication disorders with emphasis on Indian perspectives. Language in India. 9:8 August 2009. Colzato, L. S., Bajo, M. T., van den Wildenberg, W., Paolieri, D., Nieuwenhuis, S. T., La Heij, W., & Hommel, B. (2008). How does bilingualism improve executive control? A comparison of active and reactive inhibition mechanisms. Journal of Experimental Psychology: Learning, Memory, and Cognition, 34, 302-312. Coppieters, R. (1987). Competence differences between native and near native speakers. Language, 63, 544-573. Costa, A., Hernander, M., & Sabastia-Galle, N. (2008). Bilingualism aids conflict resolution: Evidence from the ANT task. Cognition, 106, 59 86. Cummins, J. (1991). Language Development and Academic Learning. In Malave, L. and Duquette, G. Language, Culture and Cognition. Clevedon: Multilingual Matters. Dash, T. & Kar, B. R. (2011). Inhibitory control mechanisms in Bilingual language processingERP Study. World Congress of Neuroscience, Florence, Italy. http://www.ibro2011.org/ Data/Files/ HtmlEditorFiles/file/pdf/

23

791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809

Dijkstra, A., & van Heuven, W. J. B. (2002). The architecture of the bilingual word recognition system: From identification to decision. Bilingualism: Language and Cognition, 5, 175197. Flege, J. E., Mackey, I. R. A., & Piske, T. (2002). Assessing bilingual dominance. Applied Psycholinguistics, 23, 567-598. Flege, J. E., Yeni-Komshian, G. H., & Liu, S. (1999) Age Constraints on Second-Language Acquisition. Journal of Memory and Language, 41, 78 104. Goodglass, H., & Kaplan, E. (1983). Boston Diagnostic Aphasia Examination (2nd ed.). Philadelphia: Lea & Febiger. Goto Butler, Y. & Hakuta, K. (2004). Bilingualism and second language acquisition. In T. Bhatia & W. Ritchie (eds.), Handbook of bilingualism. Malden MA: Blackwell Publishers. (pp. 114-145). Grosjean, F. (1998). Studying bilinguals: Methodological and conceptual issues. Bilingualism: Language and Cognition, 1, 131-149. Hakuta, K., & Suben, J. (1985). Bilingualism and cognitive development. Annual review of applied linguistics, 6, 35-45. Cambridge University Press. Hakuta, K., & Diaz, R. M., (1985). The relationship between degree of bilingualism and cognitive ability: A critical discussion and some new longitudinal data. In K. E. Nelson (Ed) Children s language, Volume 5, Hillsdale, NJ: Lawrence Erlbaum Associates.

24

810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829

Herman, D., Bongaerts, T., De Bot, K., & Schrender, R. (1998). Producing words in a foreign language: Can speakers prevent interference from their first language? Bilingualism: Language and Cognition, 1, 213-229. Jia, G., Aaronson, D., & Wu, Y. (2002). Long-term language attainment of bilingual immigrants: Predictive variables and language group differences. Applied Psycholinguistics, 23, 599 621. Johnson, J. S., & Newport, E. L. (1989). Critical period effects in second language learning: The influence of maturational state on the acquisition of English as a second language. Cognitive psychology, 21, 60-99. Kar, B. R., Khare, V. & Dash, T. (2011). Bilingualism and Cognitive Control: Is bilingualism a Cognitive Advantage? In R. K. Mishra & N. Srinivasan, Language and Cognition: State of the Art. Munich: Lincom Europa. Kertesz, A. (1982). The Western Aphasia Battery. New York: Grune & Stratton. Kroll, J. F., & de Groot, A. M. B. (2005). Handbook of bilingualism: Psycholinguistic approaches. New York, NY: Oxford University Press. Kroll, J. F., & Stewart, E. (1994). Category interference in translation and picture naming: Evidence for asymmetric connections between bilingual memory representations. Journal of Memory and Language, 33, 149-174. Kroll, J. F., Bob, S. C., Misra, M. & Guo, T. (2008). Language selection in bilingual speech: Evidence for inhibitory processes. Acta Psychologica, 128, 416 430.

25

830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848

Marian, V. (2008). Bilingual research methods. In J. Altarriba & R. R. Heredia (Eds.), An introduction to bilingualism: Principles and processes. Mahwah, NJ: Erlbaum. Marian, V., Blumenfeld, H. K., & Kaushanskaya, M. (2007). The Language Experience and Proficiency Questionnaire: Assessing language profiles in bilinguals and multilinguals. Journal of Speech Language and Hearing Research, 50, 940-967. Nordlund, I. (2005). Bilingualism- An asset to future language acquisition. C Extended essay English. Department of language and culture, Lulea University of Technology. Oyama, S. (1976). A sensitive period for the acquisition of a non native phonological system. Journal of Psycholinguistic Research, 5 (3), 261-283. Shipley, K. G., & Mcafee, J. G. (2008). Assessment in Speech-Language Pathology: A Resource Manual. Cengage Learning. Siegal, M., Iozzi, L & Lurian, S. (2009). Bilingualism and conversational understanding in young children. Cognition, 110, 115 122. Sparks, R., Patton, J., Ganschow, L., and Humbach, N. (2009). Long-term cross linguistic transfer of skills from L1 to L2. Language learning, 59:1, pp 203-243. Steven, G. (1999). Age at immigration and second language proficiency among foreign-born adults. Language in Society, 28, 555 578. Pon Subbiah (2005). Test of language proficiency: Hindi. Central Institute of Indian Languages, Mysore, India.

26

849 850 851 852 853 854 855 856 857 858 859 860 861 862

Thorum, A. R. (1986). The Fullerton language test for adolescents professional manual. Consulting Psychologists Press. Van Hell, J. G., & Tokowicz, N. (2010). Event-related brain potentials and second language learning: Syntactic processing in late L2 learners at different L2 proficiency levels. Second Language Research, 26(1), 43-74. Vasanta, D., Suvarna, A. Sireesha, J., & Bapi Raju, S. (2010). Language choice and language use patterns among Telugu-Hindi/Urdu-English speakers in Hyderabad, India. Proceedings of the International Conference on Language, Society, and Culture in Asian Contexts (PP. 57-67). Thailand: Mahasarakam University. Weber-Fox, C. M., & Neville, H. J. (1996). Maturational constraints on functional specializations for language processing: ERP and behavioural evidence in bilingual speakers. Journal of Cognitive Neuroscience, 8, 231-256. Zutell, J. & Rasinski, T. V. (1991). Training Teachers to Attend to Their Students Oral Reading Fluency: Theory Into Practice, 30, 211-217.