Determination of bilingualism subtypes and their relationship with linguistic abilities in Lebanese bilinguals

Introduction

With the growing bilingual population, bilingualism has become the new norm (Marian & Shook, 2012). Nonetheless, a standard, universal definition of bilingualism may be difficult to implement across a variety of populations and ages. Some widely accepted definitions of bilingualism have been found in the literature; however one seems to be the most relevant for the current study as a theoretical background and defines it as the ability to alternate between two languages, implying that a bilingual person “uses two or more languages in everyday life” (Grosjean, 2010). Numerous parameters and several intrinsic (age of acquisition (AoA), intensity or degree of mastery, learning context . . .) and extrinsic (language practice, context of use, social hierarchy . . .) factors influence bilingual’s skills and determine the type of bilingualism (Geiger-Jaillet, 2005; Rezzoug et al., 2007). Therefore, early bilingualism must be distinguished from late bilingualism: early simultaneous bilingualism, which implies that the individual has acquired two languages simultaneously from birth, and early sequential bilingualism, which is characterized by the acquisition of two languages very early in childhood, but successively (De Houwer, 2021; Hamers & Blanc, 1983, pp. 6–25). Bilingualism is late if the second language (L2) is learned from the age of 12 years (De Houwer, 2021; Heredia & Cieślicka, 2014, pp. 42–52). Moreover, bilingualism can be either prominent (i.e., dominant) or balanced: an individual has prominent (i.e., dominant) bilingualism when he is more proficient in a language compared with the other language, while he would be described as balanced when he is equally proficient in both languages (Birdsong, 2014; Gathercole & Thomas, 2009; Kohnert, 2008). However, in a context of relocation or immigration, where one language is socially privileged, some authors speak of the devaluation of so-called “heritage” languages (Rezzoug et al., 2007), which impacts on the practice and maintenance of this language by participants. Lambert (1974), Houwer and Ortega (2018) make a distinction between additive bilingualism, which is highly valued by the entourage and environment, and subtractive bilingualism, which is little known or devalued by society.

Therefore, today among studies targeting bilingual populations in neuroscience, many tools have been used to determine the type of bilingualism and language dominance. Self-rating questionnaires have occurred as the most applied and commonly used assessment tools. Soares and Grosjean (1984) recommend rating each language modality separately (understanding, speaking, reading, and writing), as these abilities may reveal unequal capacities. Indeed, proficiency levels in these domains are often used to determine language dominance with other factors, such as age, level of education, language acquisition context, language experience, habits, and linguistic and cultural identity (Schmid & Yılmaz, 2018). Thus, it is important to distinguish between language dominance and language proficiency, which refers to the ability to understand, speak, read, and write language (Marian & Hayakawa, 2021) and is assessed using language proficiency scores. Language dominance, however, is a construct that derives from the nature of bilingualism (Grosjean, 1998) where cross-linguistic comparisons of language proficiencies typically show linguistic dominance (where one language is found to be more proficient than the other one) (Gathercole & Thomas, 2009). Paradis (1989) added other factors that are now known to influence and shape a bilingual individual’s performance on linguistic and executive tests, such as the AoA, acquisition manner (formal or informal), form or frequency of use, literacy in each language, and so on. An important variable to consider is the AoA, which is generally used to provide the classification of early, late, or simultaneous acquisition. To quantify bilingualism while taking into consideration all the elements cited above, many questionnaires have been developed such as the Language History Questionnaire (Li et al., 2006), the Language Experience and Proficiency Questionnaire (LEAP-Q) (Marian et al., 2007), Language Background Questionnaire (Sabourin et al., 2016), Language and Social Background Questionnaire (Anderson et al., 2018), Bilingual Language Profile (BLP) (Birdsong et al., 2012), and The Contextual Linguistic Profile Questionnaire (Wigdorowitz et al., 2023). To date, the LEAP-Q exists in 36 different languages (including Modern standard Arabic), has been adapted socially and culturally for bilinguals around the world, and can be easily administered in an oral interview setting, digital or even using paper-and-pencil. It gives the opportunity to provide a full profile for bilingual participants and to determine bilingualism subtypes.

Bilingualism subtype (prominent first language [L1], balanced bilingualism, prominent L2), defined as language dominance in bilingual individuals, is generally determined using self-reported reading proficiency and self-reported speaking proficiency in the LEAP-Q (Marian et al., 2007). A significant number of studies have determined the bilingualism subtype according to raw scores on a self-reported questionnaire (Berkes et al., 2020; Lamar et al., 2019; Raji et al., 2020; Wauters & Marquardt, 2018). As a matter of fact, while using the scores collected from the BLP (Birdsong et al., 2012), Birdsong (2016) has attempted to create an index to better identify bilingualism subtype: he suggested the following formula for establishing language dominance:, (Language A Score—Language B Score) / the greater of the two scores + 1, /2. Multiplying this result by 100 will give scores ranging from 0 to 100, with 50 as the score for perfectly balanced bilingualism. Whereas other studies (Cop et al., 2015; Gasquoine et al., 2019) have calculated the proficiency ratio (L1: L2) based on various aspects of language: numerous standardized language tests to assess oral expression have been used, such as the Expressive Vocabulary Test (Williams & Wang, 1997), the Woodcock-Munoz Language Survey—Revised Picture Vocabulary subtest (Woodcock et al., 2005), the Multilingual Naming Test (MINT) (Gollan et al., 2012), and Boston Naming Test (BNT) (Kaplan et al., 1978). As a matter of fact, the relevance of naming to assess language proficiency, and more specifically the BNT, in particular, derived from the examination of aging, has been demonstrated in numerous studies (Roussel et al., 2016; Tran & Godefroy, 2015) and brain damage (Andriuta et al., 2018). Whereas, for oral comprehension, the Shipley Vocabulary Test (Shipley, 1940), Peabody Picture Vocabulary Test (Dunn & Dunn, 1997), LexTALE (Lemhöfer & Broersma, 2012), and more recently, the Ecological Momentary Assessment (Jylkkä et al., 2020) have been the most used. While other capacities seem to be influenced by language proficiency, such as articulatory rate (Kim et al., 2022) and reading speed (Cop et al., 2015), to our knowledge, no studies have adopted them as a proficiency measure in a bilingual elderly population.

Recently, two studies compared the bilingualism classification based on questionnaires (including self-rating), structured oral proficiency interviews and naming tests notably BNT and the MINT (Gollan et al., 2012; Sheng et al., 2014) and found that these four measures converged in classifying bilinguals into language-dominance groups. However, they specified that the BNT and the MINT (more specifically the BNT) classified bilinguals as more prominent English than self-rating and interview measures in their Spanish-English bilinguals (Gollan et al., 2012) and Mandarin-English bilinguals (Sheng et al., 2014). Therefore, they concluded that to have a full assessment of bilinguals’ language proficiency, a single measure is insufficient (Gollan et al., 2012; Sheng et al., 2014) and a combination of measures appears to be the best way to quantify bilingualism. Consequently, a growing number of recent studies (Cop et al., 2015; Gasquoine et al., 2019; Woumans et al., 2015) have opted for the use of a combination of subjective and objective measures on language tests obtained in both languages to determine bilingualism subtypes of their population but none confronted this combination with another language test to examine the validity of this combination as it was not the aim of their study.

This literature review shows that several factors explain inconsistencies in grouping bilinguals into subtypes such as the absence of a consistent definition of bilingualism (Anderson et al., 2018), the lack of a clear age and cutoff scores to determine the type of bilingualism (early or late; prominent, balanced), and discrepancies between the objective measurements of language performance and questionnaires. Such diversity in bilingualism classification in heterogeneous populations across studies contributes to conflicting results. Furthermore, an agreement on a single measure in the bilingual elderly neuroscience literature does not exist, despite the fact that it is recommended (De Cat et al., 2018). Therefore, a single generalizable index with a defined cutoff based on combination of subjective and objective measure is needed to capture bilingualism dominance. Its creation will facilitate direct comparisons between individuals, as well as between neuroscience research teams (Marian & Hayakawa, 2021) studying elderly population.

As regards the objective of defining an index of bilingualism, a particularly interesting country is Lebanon, as its population is currently immersed from birth in a multilingual environment. Several languages are used in this country in particular Modern Standard Arabic, Lebanese Arabic, French, English, and Armenian (Daccache et al., 2020; Hoteit, 2010; Kanaan, 2011; Makki, 2007; Saliba, 1978). The country’s official language is Modern Standard Arabic, used mainly in writing and the media, while Lebanese Arabic, used for communication purposes between individuals, is spoken by 93.7% of the Lebanese population (Leclerc, 2015). In 2007, a survey showed that 45% of Lebanese spoke French and 40% spoke English (Gingra, 2011). So, today, it is not surprising to hear Lebanese, French, and English in the same sentence (Kotob, 2002). This prevalent bilingualism in Lebanese society, described as early (Zablit & Trudeau, 2008), is also reflected in the education system: at the time of schooling, parents choose either a French-speaking or English-speaking school for their children (Kouba-Hreich & Messarra, 2020, pp. 11–26). Therefore, Lebanese children are simultaneously taught two languages (Arabic and French or Arabic and English) and two distinct alphabetical systems: the Arabic alphabet and the Latin alphabet. In 2005, the majority of pre-university education was provided by establishments following the Lebanese national curriculum, in addition to those following a French curriculum and approved or contracted by the French Ministry of Education. According to the Ministry of Education and Higher Education in Lebanon, the Lebanese national curriculum is characterized by teaching the following subjects in Arabic: Arabic, geography, history, and civic education; and the other subjects are taught in a foreign language (English or French): physics, chemistry, biology, science, foreign language (Byloun, 2015). Lebanese bilingualism (simultaneous or sequential) falls into the category of early bilingualism, which provides the opportunity to compare balanced bilingualism and prominent (i.e., dominant) bilingualism. It is important to note that even though Lebanese people live in a multilingual environment, they usually master two languages (Arabic and French or Arabic and English).

The main objective of this cross-sectional study was to take advantage of Lebanese bilingualism to establish a multidimensional bilingualism index and assess its relationship with a confrontation naming test. The secondary objective was to assess the relationship between the resulting bilingualism subtypes with other measures of language and general cognitive screening. Our hypothesis was that the resulting bilinguals’ subgroups will differ on other linguistic and screening tests.

Materials and methods

Results

First, we will expose the results of the correlation analysis to establish a multidimensional bilingualism index then we will explore its relationship with a confrontation naming test. Second, we will investigate the relationship between the bilingualism subtype created from the bilingualism index with other linguistic abilities (articulatory rate, oral comprehension, reading) and general cognitive screening.

Bilingualism subtype

The correlation analysis showed that all nine LEAP-Q items (Supplemental Table) were associated with the differential Ar-Fr dBNT score (Table 1). By stepwise linear regression, the best account of the Ar-Fr dBNT score was provided by the LEAP-Q combined score of SpeakLev Ar-Fr and UndersLev Ar-Fr (R² = .435, p = .0001). Both SpeakLev Ar-Fr and UndersLev Ar-Fr scores were grouped using principal component analysis to create the bilingualism index.

OPEN IN VIEWER

Table 1.

Correlations between the nine LEAP-Q items and Arabic-French differences on Boston Naming Test score (Ar-Fr dBNT score).

Ar-Fr score	Ar-Fr dBNT score	p
Language dominance	−0.56	0.001
Acquisition order	−0.52	0.001
Global self-reported proficiency	−0.074	0.23
Age started learning	−0.35	0.001
Became orally fluent	−0.4	0.001
Started reading	−0.42	0.001
Became fluent reading	−0.41	0.001
Speaking and understanding self-reported proficiency	0.66	0.001
Reading self-reported proficiency	0.6	0.001

Note. LEAP-Q: Language Experience and Proficiency Questionnaire; Ar-Fr dBNT score: differential Arabic-French score on the Boston Naming Test.

Bold: LEAP-Q combined score of Speaking and understanding self-reported proficiency that best account of the Ar-Fr dBNT score.

ROC curve analysis of the bilingualism index with the dichotomized Ar-Fr dBNT score as the state variable showed a high area under the curve (0.893, 95% CI = [0.823, 0.962], p = .0001) (Supplemental Figure). The cutoff scores of the bilingualism index were determined (Figure 1): a cutoff score < −2 provided the best discrimination to identify participants with higher naming performance in French (i.e., negative Ar-Fr dBNT score) and therefore they belonged in the prominent French subgroup; a cutoff score > 2, selected participants with higher naming performance in Arabic (i.e., positive Ar-Fr dBNT score) and therefore creating the prominent Arabic subgroup. Finally, participants who scored between −2 and 2 were considered as balanced bilinguals. On this basis, the group was categorized as prominent Arabic (N = 35), balanced bilingualism (N = 58), and prominent French (N = 7) (Table 2). The three subgroups did not differ according to age, education level, sex, handedness, or order of completion (Table 2).

OPEN IN VIEWER

Figure 1.

Relationship (linear regression line with 95% confidence interval) between the differential Arabic minus French Boston Naming Test score (Naming_Sc_Ar_Fr) (Y axis) and the bilingualism index (X axis).

OPEN IN VIEWER

Table 2.

Characteristics of the three bilingualism subgroups expressed as the number (n) of participants and the mean ± standard deviation.

	All	Prominent Arabic	Balanced	Prominent French	p
N	100	35	58	7
Sex (women/men) (n)	58/42	18/17	36/22	4/3	0.6
Age	68.2 ± 9.40	67.60 ± 9.12	68.22 ± 9.49	71.11 ± 10.49	0.7
Education level (n)					0.5
Level 1	28	7	18	3
Level 2	21	10	10	1
Level 3	51	18	30	3
Handedness (n)					0.6
Left-handed	7	3	3	1
Right-handed	93	32	55	6
Order of completion (n)					0.9
Arabic tests first	53	19	31	3
French test first	47	16	27	4
Boston Naming Test (/34)
Arabic	21.00 ± 4.39	22.20 ± 3.90	20.50 ± 4.67	19.28 ± 3.15
French	14.92 ± 7.02	9.77 ± 3.39	16.80 ± 6.70	25.14 ± 2.69
Articulatory rate (syllable/s)
Arabic	5.10 ± 1.31	5.37 ± 1.24	5.01 ± 1.34	4.53 ± 1.34
French	4.41 ± 1.24	4.12 ± 0.92	4.51 ± 1.36	5.05 ± 1.33
Token test (/36)
Arabic	33.97 ± 1.98	33.87 ± 1.95	34.25 ± 1.90	33.78 ± 2.80
French	33.37 ± 2.40	32.44 ± 2.50	33.45 ± 2.20	34.21 ± 2.90
Rate of syllable reading(number/s)
Arabic	0.76 ± 0.03	0.87 ± 0.04	0.83 ± 0.03	0.57 ± 0.08
French	1.86 ± 0.06	1.80 ± 0.08	1.91 ± 0.06	1.86 ± 0.17
Mini-mental state examination (/30)
Arabic	27.90 ± 2.07	27.89 ± 2.00	28.00 ± 2.06	27.71 ± 2.75
French	27.30 ± 2.50	26.51 ± 2.35	27.71 ± 2.64	28.14 ± 2.04

Note. Education level: Level 1: ⩽ 8 years; Level 2: 8–11 years; Level 3: ⩾ 12 years.

As expected, BNT scores differed according to (1) bilingualism subtype (p = .001, η² = 0.19) due to a better score for the prominent French subgroup (22.4 ± 1.4) than the balanced subgroup (18.7 ± 0.47), which was, in turn, better than that of the prominent Arabic subgroup (16.11 ± 0.6), and (2) according to language (p = .001, η² = 0.11), due to an overall better score in Arabic than French (completion order: p = .6) (Table 2). Consistent with the method of subgroup determination, the bilingualism subtype × language interaction was significant (p = .0001, η² = 0.381) due to a higher score in Arabic for the prominent Arabic subgroup and a higher score in French for the prominent French subgroup.

These analyses show that two scores from the LEAP (SpeakLev Ar-Fr and UndersLev Ar-Fr) can provide a bilingualism index that allows us to trichotomize the study population into three bilingualism subtypes consistently with scores on the BNT in both languages. The validity of this index was now examined on other tests assessing language skills and screening of cognitive abilities.

Discussion

This study aimed to re-examine the methods of determination of bilingualism subtype by establish a multidimensional bilingualism index and assess its validity with language measures and general cognitive screening. Our results yielded that linguistic abilities of bilinguals defined according to selected indices from the LEAP-Q are influenced by the language of assessment in the expected direction, with an advantage conferred by the prominent language for naming, articulatory rate, oral comprehension, and speed of syllable reading. In addition, performance was, overall, slightly better in Arabic for oral comprehension and articulatory rate, whereas the reverse was observed for reading speed. Finally, the advantage conferred by the prominent language was also observed for the MMSE, a screening test with high linguistic demands. Our findings converge with some authors (Cop et al., 2015; Gollan et al., 2007, 2012; Kim et al., 2022; Marian et al., 2007; Milman et al., 2018; Sheng et al., 2014; Sheppard et al., 2016) and diverges on some points with others (Gollan et al., 2012; Sheng et al., 2014).

Confrontation between self-rating and actual linguistic performance showed that two items of the LEAP-Q, namely the proficiency level in speaking and understanding, were associated with the differential of naming between languages. The high area under the ROC curve indicates a close relationship between self-rated and actual measures of bilingualism on the BNT, thus showing that these two LEAP-Q items capture most of the bilingualism subtype. This is supported by the finding that the addition of other LEAP-Q items in the regression analysis did not meaningfully improve the explained variance. These results indicate that the use of these two LEAP-Q items may be especially relevant to classify bilingualism subtype. The actual measure of linguistic abilities between languages was based on a confrontation naming test performed in both languages; this measure draws on a considerable number of studies showing the relevance of naming for assessing language skills, particularly those derived from the study of aging (Roussel et al., 2016; Tran & Godefroy, 2015) and brain damage (e.g., Andriuta et al., 2018; Godefroy et al., 2002). As a matter of fact, our results are consistent with other findings with different subjective measures. Notable with Gollan et al. (2012) findings in Spanish-English bilinguals on the classification convergence between the BNT and the Language History Questionnaire and with Sheng et al. (2014) in Mandarin-English bilinguals on the classification convergence between the BNT and self-ratings. However, in both studies the MINT indicated higher degree of English dominance compared with other measures and even higher for the BNT in Gollan et al. (2012).

Importantly, our study also showed a close relationship of our bilingualism classification with other linguistic abilities, namely articulatory rate, oral comprehension, reading speed, and cognition, based on a screening test with high linguistic demands (as it includes oral comprehension, oral expression, repetition, reading, and writing subtests). This supports the external validity of our bilingualism classification. Our finding converges with Marian et al. (2007) outcome of a correlation in the expected direction between self-reported understanding, speaking, and reading of the LEAP-Q and performances on tests assessing naming, oral comprehension, reading, written comprehension, and grammaticality judgments. Therefore, we can recommend the use of our bilingualism index to the assessment of bilingual population aged 55 years and over, in the perspective of cognitive testing. By simply calculating the average of SpeakLev Ar-Fr and UndersLev Ar-Fr, authors can quantify the language dominance of their healthy elderly population and pathological older adults with mild to moderate cognitive impairment. This can be very helpful for clinicians opting for linguistic and cognitive assessment in multilinguals countries such as Lebanon.

Our analyses revealed significant differences in BNT scores between bilingual subgroups, with an advantage observed in the prominent language. A higher level of proficiency implies fewer naming errors and a richer lexical stock, which could, in turn, explain the higher scores in Arabic for the prominent Arabic subgroup and inversely in French. Two studies (Gollan et al., 2007; Sheppard et al., 2016) examined performance on the BNT naming test in both languages of bilinguals. Our results support the findings of Gollan et al. (2007) on 29 English-Spanish balanced and unbalanced bilinguals who performed better in their prominent language. Furthermore, in the balanced subgroup, pictures with cognate names were named more easily and accurately across languages (Gollan et al., 2007). It is important to note that the cognate effect, generally observed in naming tests of bilinguals, was not truly relevant in our study, as Lebanese Arabic and French do not have many cognate names and more precisely on the list of words included in the BNT. Moreover, our Arabic-French balanced bilinguals performed numerically better in Arabic than French, a finding also observed by Sheppard et al. (2016), who found better performance in English than French among balanced French-English bilinguals. This finding can be explained by several mechanisms. First, individuals in the balanced subgroup may be more exposed to one language in their daily life. As a matter of fact, our participants were more frequently exposed to Arabic than French in their daily life. Second, the difficulty of BNT items may differ between languages; however, we were not able to control this factor as a frequency dictionary in Lebanese Arabic does not exist yet. Thus, the hypothesis of differences in the difficulty of the BNT between languages is yet to be generally proven, in particular, between Lebanese Arabic and French.

A faster articulatory rate was found in the prominent language. This result converges with the observation of Kim et al. (2022) of a significantly slower speaking rate in L2 English speakers than L1 English speakers and extends it to Arabic-French bilinguals. This finding can be explained by the fact that word production in the L2 is less automatic; therefore, lexical retrieval and articulatory processing require the recruitment of additional neural networks compared with the one involved in the L1 (Liu et al., 2010). The overall faster rate in Arabic was due to the small size of the French prominent subgroup.

The linguistic advantage in the expected direction was also observed for the shortened Token test. This provides an objective measure of the better oral comprehension skills conferred by the prominent language, which has not been previously documented by a sentence comprehension test, in particular, for Arabic-French bilinguals. The overall higher score in Arabic was due to the small size of the prominent French subgroup.

Reading speed differed across bilingualism subtypes, with an advantage in the prominent language. This advantage in the prominent language supports the findings of Cop et al. (2015) who defined L1 and L2 reading patterns in bilinguals: in comparison to L1 readers, L2 readers take 20% more time to read sentences, fixe words 21% longer, engage in 12% shorter saccades, and skip 4.6% less words. This difference in speed between subtypes suggests that language proficiency may play a crucial role on reading outcomes. In addition, reading speed was globally faster in French than Arabic. This is the sole test in which such a pattern was observed and contrasts sharply with the overall faster articulation rate in Arabic. The selected color names are monosyllabic in French and disyllabic in Arabic. However, this is unlikely to explain this difference, as we analyzed the rate of syllable reading. The contribution of different senses of visual scanning (left to right in French, right to left in Arabic) between languages constitutes a potential explanation that would require a specific study.

We found the expected advantage in the prominent language for the MMSE. These findings support those of earlier studies showing an intimate relationship between MMSE scores and language proficiency (reported from standardized tests), with language proficiency accounting for 67% of MMSE variance in a bilingual Spanish–English group and 42% of the variance in a bilingual Asian Indian-English group (Milman et al., 2018). This supports the heavy reliance of the MMSE on language components (Tombaugh & McIntyre, 1992).

Limitations

This study had several limitations. First, we used one naming test as the main measure to assess oral language capacities. Although we adapted it to the Lebanese culture, its use can be debated, as item difficulties (frequency, linguistic characteristics of names, etc.) were not assessed between languages. Therefore, it would be informative to compare different naming tests created specifically for multilinguals on a large sample of an elderly population. However, this test was found to appropriately assess controls and patients from diverse cultures and languages throughout the world (Lo et al., 2019) and, more importantly, to show that naming relies on a common brain mechanism (Weaver et al., 2021). Despite this potential shortcoming, a strength of our study was the large sample of bilinguals assessed in both languages and the variety of native bilingualism subtypes in the Lebanese population. Furthermore, we utilized easily accessible tests for professionals, as well as a user-friendly scale from the LEAP-Q for participants. In particular, the bilingualism index that we created appeared to be associated with performance for both languages in several linguistic tests that assess the main aspects of oral and written language, as well as screening of general cognitive abilities. Second, Marian and Hayakawa (2021) recommended “the use of large, diverse datasets and advanced statistical methods to select and weight attributes to implement flexible methods of administering, weighting, and scoring variable standards of comparison based on demographic characteristics to determine relative bilingualism” and its change over time and to establish norms. In our study, many elements were taken into consideration as we followed Kremin and Byers-Heinlein (2021) suggestion that bilingualism measurement should follow either categorical or continuous models depending on the relation between a study hypothesis and its observable measures: in clinical neuroscience, and more precisely in studies concerning one time measurement, we find categorical model more practical, efficient and best suited for the addressed questions. Therefore, certain requirements proposed by Marian and Hayakawa (2021), such as assessing changes in bilingualism over time could not be taken into consideration in the categorical model in our study. Adding a number of key questions and methods to incorporate all these variables will be highly challenging but necessary.

Finally, our results demonstrated that a simple index can accurately characterize adult bilingualism subtypes and therefore has clinical implications. As a matter of fact, health care professionals increasingly encounter bilingual patients, especially in Lebanon where a majority of the population is bilingual, and a variety of bilingualism subtypes exists, a circumstance that frequently prompts inquiry into the influence of language choice on the assessment of linguistic and cognitive abilities, and consequently, the determination of the cognitive status required for diagnosis. Our index offers clinicians an easy and fast tool compared with the usual procedure used to determine individuals and patient’s bilingualism subtype. More broadly, the index’s validity in other bilingual populations is warranted for generalizability of the present findings.

Supplemental Material