This study investigates bilingual performance on the English and Spanish Boston Naming Tests (BNTs) while controlling for object familiarity and U.S. acculturation. Previous studies suggest that bilingualism negatively affects naming skill; however, object familiarity, which may be culturally influenced, and U.S. acculturation level have not been formally investigated. The current sample comprised 74 well-acculturated bilinguals and 52 English monolinguals. Participants judged their familiarity with BNT objects and later named the objects in either English or Spanish. Both groups rated BNT objects to be comparably familiar. However, bilinguals underperformed relative to monolinguals. In fact, those bilinguals born and raised in the USA and educated solely under English instruction were unable to match monolinguals' superior naming performance. These results underscore a language disadvantage in naming even for native-born, highly acculturated, English proficient bilinguals and suggest that the BNT is language specific and perhaps unsuitable for testing bilingual populations.
The ability to name objects is an essential component of effective communication. A reduction or total loss in naming ability is one of the most frequent signs of cognitive–linguistic dysfunction and can emerge early in the course of Alzheimer's disease, after left cerebral hemisphere stroke, traumatic brain injury (Henderson, Frank, Pigatt, Abramson, & Houston, 1998), or as a result of temporal lobe epilepsy. Although naming tasks play an important role in assessing and diagnosing many pathological conditions, the Boston Naming Test (BNT; Kaplan, Goodglass, & Weintraub, 1983) is one of the few standardized tests available to assess confrontation naming skill. The BNT consists of 60 line drawings of different objects, arranged in an order of increasing naming difficulty. Standard administration is discontinued once a person commits eight consecutive naming errors, with the assumption that the examinee is not likely to correctly name subsequent, more difficult items. Further background and test description are provided by Lezak, Howieson, and Loring (2004).
Various factors have been found to affect naming skill, including familiarity, word frequency (i.e., frequency with which a word appears in print), and age-of-acquisition (AoA; i.e., age at which one acquires a given word). For example, Ferraro, Blaine, Flaig, and Bradford (1998) found that the more familiar an object was to a person or the more frequently the word appeared in print, the faster the person was able to identify the word as real. Word frequency may affect the speed and accuracy with which a person performs picture naming tasks (Jescheniak & Levelt, 1994; Oldfield & Wingfield, 1964, 1965), and while some authors contend that AoA has a stronger influence on naming skill than word frequency, others believe that the two variables influence naming skill differently and possibly interact. For example, Morrison, Ellis, and Quinlan (1992) reanalyzed data from the Oldfield and Wingfield experiment to include AoA as a variable and found that naming speed correlated with frequency, but the association with AoA was more consistent and stronger. In an analysis of over 140 studies on AoA research, Juhasz (2005) reports that, when AoA is included in naming studies, its effects on naming skill consistently reach significance, whereas word frequency effects are less conclusive. Barry, Morrison, and Ellis (1997) studied the effects of familiarity, frequency, and AoA on picture naming. While naming latency correlated with all variables, an interaction between frequency and AoA emerged. Specifically, high-frequency names were generated faster than low-frequency names but only for late-acquired words; frequency had no effect on early-acquired names.
While familiarity, word frequency, and AoA affect picture naming skill, culture and language are important to consider. An examinee's ability to name an object is contingent upon a sufficient level of familiarity with that object and its name, and people of different cultures likely encounter objects at different frequencies. Object names are also likely to differ in frequency and use across languages and, within a given language, certain objects might have more than one accepted name. Although a discussion of confrontation naming in non-English languages is beyond the scope of this paper, it is noteworthy that the first normative study on the Spanish BNT (Kaplan, Goodglass, & Weintraub, 1986) reflects such variability in naming (Allegri et al., 1997). Although the Spanish BNT replaces various drawings to address language and cultural discrepancies, the order of presentation and the norms published with the test are identical to its English counterpart. Allegri and colleagues concluded that the current order of presentation does not reflect graded naming difficulty level for the Spanish-speaking population they evaluated. A similar finding might also emerge with certain English-speaking populations. Given the standard administration procedures, this could pose significant problems in interpreting test results.
Knowledge of more than one language may also play a critical role in picture naming and is intimately related to language proficiency. The ability to speak more than one language is common in most parts of the world and is becoming increasingly common in the United States. Bilinguals typically demonstrate higher proficiency in one language compared with the other, whereas balanced bilinguals—bilinguals who speak both languages with native-like proficiency—are rare (Gollan, Montoya, Fennema-Notestine, & Morris, 2005). Studies indicate that achieving and maintaining proficiency in two languages comes at a cost to the dominant language, regardless of whether the dominant language is the first or second language acquired (Michael & Gollan, 2005; Ransdell & Fischler, 1987). Bilingual adults demonstrate weaker verbal skills in each language compared with monolingual speakers of those languages, with weaknesses evident in receptive vocabulary size as well as slowed and less-accurate verbal expression on picture naming and fluency tasks (Bialystok, Craik, & Luk, 2012). Bilingual performance on the BNT confirms these findings. Kohnert, Hernandez, and Bates (1998) report preliminary normative data on 100 young adult Spanish–English bilinguals on the English BNT. The test was administered in both English and Spanish to each participant, with an order of presentation counterbalanced across participants. Although participants performed better in English than in Spanish, English performance was well below published norms. Kohnert and colleagues suggest that linguistic, cultural, and experiential variables explain underperformance among their cohort. Roberts, Garcia, and Desrocher (2002) investigated BNT performance of two bilingual groups (Spanish–English and French–English) against an English monolingual control group. Both bilingual groups performed significantly worse than their monolingual counterparts. However, the two bilingual groups achieved comparable mean scores despite their different cultural backgrounds. The authors conclude that low performance among bilingual groups cannot be attributed to cultural differences alone. Gollan, Fennema-Notestine, Montoya, and Jernigan (2007) investigated whether awarding credit for naming objects in either English or Spanish might improve performance outcomes for their bilingual older adult sample. Balanced bilinguals named fewer pictures in their dominant language compared with unbalanced bilinguals, and balanced bilinguals benefited from either-language scoring, whereas unbalanced bilinguals did not.
Although bilinguals tend to underperform on the BNT compared with monolinguals, several factors may play a role in underperformance, including culture, level of acculturation, and object familiarity within that culture. To our knowledge, none of the studies mentioned earlier assessed participants' acculturation level or level of familiarity with BNT objects. Strong BNT performance is contingent upon sufficient familiarity with the objects presented. If familiarity with BNT objects is culturally mediated, then this could at least partially explain why less-acculturated bilinguals underperform on the test. The aim of this study is to assess the contribution of acculturation and object familiarity on BNT performance in a bilingual population. The present study includes a measure of familiarity and acculturation to examine linguistic, cultural, and experiential variables that might account for performance differences between language groups. We further aimed to measure how well highly acculturated bilinguals born and raised in the United States perform on the BNT compared with their monolingual counterparts. In clinical settings, highly acculturated bilingual subjects would undoubtedly receive the English BNT over the Spanish version, given their level of English proficiency. Understanding how this population performs on the BNT may allow for more accurate interpretation of test results.
Given the expected variability of acculturation level within our bilingual cohort, bilinguals were expected to report less familiarity with BNT objects compared with monolinguals, indicating that less-acculturated bilinguals encounter BNT objects less often. In terms of naming skill, bilinguals were expected to name BNT objects better in English than in Spanish. Overall, monolinguals were expected to outperform bilinguals.
Miami is a vibrant, multicultural city with a large, diverse Hispanic community. Because the current study focuses on BNT performance among Spanish–English bilinguals, the inclusion of participants from various Hispanic cultures was welcomed. A total of 183 participants were recruited from colleges and select communities within South Florida. Of those who participated, 126 met criteria for inclusion. Specifically, participants were required to be between 19 and 54 years of age, to have completed at least 13 years of education, to have normal or corrected-to-normal vision and normal hearing, and to have no history of neurological disorder (e.g., stroke, seizures, transient ischemic attacks, reading disability/dyslexia, language disabilities), traumatic brain injury, substance abuse, or psychiatric illness (e.g., depression). Participants were required to be either English monolingual or Spanish–English bilingual. Participants were deemed bilingual if they identified themselves as such on a self-report measure, were raised in a household where Spanish speaking was prevalent, and reported having learned either Spanish as a first language or Spanish and English simultaneously. To ensure substantial English language exposure and proficiency, bilinguals were required to have acquired English by 10 years of age. Participants were classified as monolingual if they identified themselves as such on a self-report measure and were raised in an English monolingual household. The CAU Institutional Review Board approved all procedures, and participants provided written consent for their participation.
Materials and Procedures
Because some bilingual participants would ultimately be assigned to name BNT objects in Spanish, test materials were created by combining all drawings appearing in the English and Spanish BNTs. The Test de Vocabulario de Boston, or Spanish BNT, was first published in 1986 (Kaplan et al., 1986). Although the first edition of the Spanish BNT maintains 48 of the original 60 drawings and replaces 12, the latest Spanish revision (2nd ed., 2005) maintains 53 of the original drawings and replaces 7. For instance, the item pretzel, which has no Spanish-equivalent name, is replaced with a drawing of a cupcake, and therefore, both would be included in the study.
Participants were evaluated in a well-lit, noise-free environment. Evaluations were conducted in groups, except for four participants who were tested individually due to scheduling conflicts. To ensure an unbiased assessment of familiarity, participants were first asked to judge their level of familiarity with each of the BNT items without knowledge that they would later be asked to name them. “Familiarity packets” contained images of all BNT stimulus items (60 English BNT items and 12 items specific to the Spanish BNT), arranged eight per page, and subjects were instructed to rate their familiarity with each drawing using a Likert-type scale provided below each drawing (1 = completely unfamiliar; 9 = highly familiar). Next, participants completed a questionnaire on their language use and proficiency. Specifically, participants were asked to self-classify as monolingual, bilingual, or other; rate their level of proficiency within each language across the spectrum of speech, comprehension, reading, and writing using a Likert-type scale (0 = no knowledge; 5 = moderate ability compared with native speaker; 10 = native speaker level of ability); and estimate the ratio each language is used in various life situations (e.g., at work, home, school, with friends, while watching TV/movies, and reading). Based on the participant's self-classification as monolingual or bilingual, he or she was assigned to name the BNT items in one designated language. Monolinguals received naming packets instructing them to name objects in English, whereas bilinguals were randomly assigned packets that either instructed to name objects in English or Spanish. “Naming packets” were similar to familiarity packets, except the rating scale was replaced with a blank line on which respondents were instructed to write the name of each item. Finally, participants completed a background questionnaire and the Short Acculturation Scale for Hispanics (SASH; Marín, Sabogal, Marín, Otero-Sabogal, & Perez-Stable, 1987).
The SASH is a brief acculturation scale developed to assess acculturation across various Hispanic groups. The SASH consists of 12 items that form three subscales—Language Use, Media, and Ethnic Social Relations—which together correlate highly with various factors, including respondents' generation, length of residence in the United States, age at arrival, and ethnic self-identification (Marín et al., 1987). Items provide five answer choices arranged in Likert-type format ranging from 1 (“only Spanish” or “all Latinos/Hispanics”) to 5 (“only English” or “all Americans”). At test construction, the 12 items of the SASH showed similar factor structures in a group of 363 Hispanics of Mexican, Cuban, and Central-American descent, and 228 non-Hispanics (Marín et al., 1987). The first five questions of the scale load onto “Language Use” and consist of questions measuring the respondent's preference for speaking a given language in a number of settings, such as during childhood, at home, at school/work, with friends, and while thinking. The next three questions load heavily on the “Media” factor and tap into the respondent's preference and use of English/Spanish language media. The last four items form the “Ethnic Social Relations” factor, which assesses the preferred ethnicity of those with whom the respondent interacts (e.g., Latinos/Hispanics vs. Americans). The Alpha coefficient was 0.92 for the total 12 items, 0.90 for “Language,” 0.86 for “Media,” and 0.98 for “Ethnic Social Relations” (Marín et al., 1987). The respondent's total score on the 12 items had a significant correlation to the person's generational level (0.65, p < .001), with scores on Language items correlating at 0.43, p < .01, with this same generational estimate. The total score also correlated at 0.70, p < .001, with the proportion of life spent residing in the United States, estimated as the respondent's length of residence in the United States divided by his or her actual age (Marín et al., 1987). The 12-item scale was found to discriminate between first- and second-generation Hispanics, t(303) = 13.74, p < .001. First-generation Hispanics scored a mean of 2.37, second-generation Hispanics scored a mean of 3.42, and non-Hispanics scored a mean of 4.63. Marín and colleagues suggest a cutoff score of 3.0 to distinguish those who are more acculturated from those who are less (Posner, Stewart, Marín, & Perez-Stable, 2001).
Of the 126 participants that comprised the final sample, 74 were bilingual and 52 were monolingual. Bilinguals were equally divided by object naming assignment: 37 bilinguals assigned to name the BNT items in English (BE group) and 37 bilinguals assigned to name the items in Spanish (BS group). Women comprised 64.9% of the BE group, 83.8% of the BS group, and 61.5% of the monolingual group (ME). The mean age for the overall sample was 30.6 years (SD = 8.45). An observed difference in age between the groups emerged, F(2, 123) = 5.44, p < .01, with monolinguals (M = 33.4, SD = 8.37) being older than the BS group (M = 28, SD = 8.86); however, the BE group (M = 29.27, SD = 7.06) did not significantly differ in age from the ME or BS groups. Total years of education for the BS group (M = 15.81, SD = 2.17, range 13–20) was significantly lower than for the BE (M = 17.11, SD = 2.34, range 13–21) and ME (M = 16.98, SD = 2.23, range 13–22) groups, F(2, 123) = 3.91, p = .02. Educational level between the BE and ME groups was not significantly different.
Bilinguals reported high exposure to U.S. culture overall. Half were born and raised in the United States. Most bilinguals had lived in the United States an average of 23.63 years, with all but three having lived in the United States for 10 years or more. Bilinguals had an average of 14.66 years of education under English instruction. On the SASH, bilinguals collectively scored a mean of 3.32 (SD = 0.43). Separately, BE scored a mean of 3.40 (SD = 0.48) and BS scored a mean of 3.24 (SD = 0.38). Based on the 3.0 cutoff score proposed by Marín and colleagues, these mean scores indicate that the bilingual sample was well acculturated (Posner et al., 2001).
Mean familiarity ratings were calculated for all subjects (N = 126) on the 72 BNT objects. Mean familiarity scores for the ME, BE, and BS groups were 7.1 (SD = 1.8), 7.4 (SD = 1.6), and 7.2 (SD = 1.5), respectively. A one-way ANCOVA, with age and education as covariates, revealed no significant difference between groups, F(2, 121) = .67, n.s., suggesting that participants were comparably familiar with BNT objects.
Object Naming—Three-Group Comparison
To allow for performance comparisons across language groups, an object naming score was calculated for each participant based on the 53 items common to both the English and Spanish versions of the BNT. Naming responses received credit if they matched those published in the English (Kaplan, Goodglass, & Weintraub, 2001) or Spanish (Kaplan, Goodglass, & Weintraub, 2005) BNT scoring booklet, with spelling judged leniently. An ANCOVA, with age and education as covariates, revealed a significant difference in naming skill across the three groups. Levene's test of equality of error variance was significant (p = .009); therefore, based on recommendation by Keppel and Wickens (2004), we set a more stringent rejection criterion (from p < .05 to p < .01), and our results were still significant, F(2, 121) = 84.001, p < .001, ηp2 = .581. The ME group outperformed both the BE and BS groups, and the BE group outperformed the BS group (ME: M = 46.69, SD = 3.61; BE: M = 42.49, SD = 6.97; BS: M = 29.51, SD = 6.95).
Object Naming—Two-group Comparison
The following analyses were conducted on a subsample of the BE and ME groups. Specifically, only participants born and raised in the United States and educated solely under English instruction were included in the following analyses (BE: n = 19; ME: n = 48). More than half of these bilingual participants acquired English from birth, and all acquired English by age 6 years. Although the BE participants were raised in bilingual households, they were exposed to similar cultural and educational systems as the monolinguals and were English dominant. On a self-rating scale of language proficiency on which bilingual participants were asked to rate their language skills compared with a native speaker (Likert-type scale range: 0–10; anchor descriptors, 0 = no knowledge; 5 = moderate ability compared to native speaker; 10 = native speaker level of ability), these bilinguals reported being highly proficient in English and judged their language skills to be better developed in English compared with Spanish (English: M = 9.43, SD = 1.11; Spanish: M = 8.50, SD = 1.41).
The mean age of the ME and BE subgroups was 33.25 and 28.26 years, respectively, t(65) = 2.44, p = .02, η2 = .08. The ratio of men to women was similar between groups (ME: 39.6% men and 60.4% women; BE: 36.8% men and 63.2% women), χ2 (1) = .043, n.s. All participants had at least 13 years of education, and the level of education did not differ significantly between groups, t(65) = .249, n.s.
A one-way ANCOVA, with age serving as the covariate, revealed no significant difference between group mean familiarity ratings on the 60 original BNT items (ME: M = 7.0, SD = 1.89; BE: M = 7.78, SD = 1.48), F(1, 64) = 3.70, n.s. Regarding naming outcome, lenient and strict naming scores were calculated for each participant. Given that all respondents had named objects in English, strict scoring required naming responses to match those published in the English BNT (Kaplan et al., 2001). Lenient scoring awarded credit for synonyms, spontaneous Spanish responses, significant misspellings, and word truncations (e.g., naming “rhino” for rhinoceros). Multivariate analysis of covariance (MANCOVA), with age serving as the covariate, revealed a significant disadvantage in naming skill for bilinguals across all scoring conditions [see Fig. 1; lenient scoring F(1, 64) = 7.74, p = .007, ηp2 = .11; strict scoring F(1, 64) = 6.11, p = .02, ηp2 = .09]. Thus, although both groups had lived in the United States since birth and had been educated fully under English instruction, bilinguals continued to perform significantly below their English monolingual counterparts, regardless of scoring leniency.
To further explore naming performance between the BE and ME subgroups, item difficulty was calculated on each of the 60 BNT items using lenient scoring results (see Table 1). Due to the limited number of participants, no test statistics were calculated, but the pattern of item difficulty suggests that monolingual participants performed at or near ceiling level until the last 10 items of the test. In comparison, bilinguals evidenced greater naming variance beginning with item 34, and naming pattern was more erratic. Table 1 reflects that BNT items are more challenging to name for bilinguals, and the standard order of presentation is not appropriately rank-ordered for this population.
|Object||Item difficulty||Object||Item difficulty|
|Object||Item difficulty||Object||Item difficulty|
Note: BNT = Boston Naming Test; Monolingual English (n = 48); Bilingual English (n = 19); Objects listed in order of presentation.
In a matched group analysis, the 19 U.S. born bilingual participants were matched to 19 monolingual participants based on age, years of schooling, and highest level of parental education. The mean age of the BE and ME matched subgroups was 28.26 (SD = 5.44) and 28.11 (SD = 5.08) years, respectively, t(36) = .093, n.s. Years of schooling averaged 17.21 (SD = 2.12) years for the BE group and 17.37 (SD = 2.17) years for the ME group, t(36) = .227, n.s. Participants were also matched on parental education, with the highest reported level of education achieved by either parent (fewer than 12 years; high school diploma; some college; undergraduate degree; post-graduate degree) serving as the match criterion. Sex was balanced across groups, χ2(1) = .11, n.s.
The matched groups were equally familiar with the 60 English BNT objects (BE: M = 7.78, SD = 1.48; ME: M = 6.87, SD = 1.99), t(36) = 1.60, n.s. We further compared the two groups on familiarity ratings for each of the 60 items. In all but one case, the BE group rated their object familiarity either the same as or higher than the ME group. In the one case, the difference between the groups did not reach statistical significance. Nevertheless, a significant disadvantage in naming skill persisted for bilinguals across strict and lenient scoring conditions. Under strict scoring, the BE group achieved a mean score of 47.84 (SD = 4.73) and the ME group achieved a mean score of 51.26 (SD = 4.28), t(36) = 2.34, p = .025, Cohen's d = 0.76. Under lenient scoring, the BE group achieved a mean score of 48.26 (SD = 4.63) and the ME group achieved a mean score of 51.95 (SD = 4.1), t(36) = 2.60, p = .014, Cohen's d = 0.84.
The current study explores object naming performance discrepancies between Spanish–English bilinguals and English monolinguals. While various factors have been shown to affect picture naming skill (e.g., word frequency, AoA), culture and language proficiency play an important role and necessitate special attention when assessing multilingual groups. Previous studies have shown that bilingual groups underperform on the BNT compared with English monolingual cohorts, but the degree to which culture and experiential variables influence such findings is unclear. An examinee's ability to name an object relies on a sufficient level of familiarity with that object and its name. Given the likelihood that objects are encountered with varying frequencies across cultures, the incorporation of an object familiarity rating scale and an acculturation scale in this investigation provides an opportunity to assess cultural and experiential influences on BNT performance. Based on anticipated variability in acculturation level within our bilingual sample, familiarity ratings were expected to show bilinguals to be less familiar with BNT objects than monolinguals due to cultural bias. This hypothesis, however, was not confirmed: bilinguals and monolinguals judged BNT objects to be equally familiar. This finding is likely due to the bilingual group's high level of acculturation, supported by the acculturation scale (SASH) results and demographic information (i.e., 50% of bilingual participants were born and raised in the United States, with most having lived in this country an average of 23.63 years). If object familiarity is culturally mediated, well-acculturated bilinguals, and their monolingual counterparts would be expected to judge BNT objects to be equally familiar. Comparable familiarity ratings, however, did not translate into equivalent naming skill.
As expected, bilinguals named BNT items significantly better in English than in Spanish. Although the BS group was younger and slightly less educated than the other groups, poor Spanish-naming performance suggests that bilinguals were English dominant overall. Stronger English naming skill is consistent with this group's high level of U.S. acculturation and substantial exposure to education under English instruction. Nevertheless, under the English naming condition, bilinguals underperformed compared with monolinguals, and this finding persisted even for a subgroup of bilinguals born and raised in the United States and educated solely under English instruction. This latter finding has significant implications for interpreting BNT performance results in a clinical setting. Given that these bilinguals had lived in the United States for their entire lives, were well educated, and considered themselves highly proficient in the English language, they may be expected to perform like monolinguals on the BNT. Yet, highly acculturated bilinguals were weaker on picture naming compared with monolinguals, even when matched by age, years of education, and parental educational level, suggesting that bilingualism itself dampens confrontation naming skill.
Item difficulty analysis (Table 1) reflects the increased variability in naming skill in the current bilingual sample. Whereas monolinguals performed at or near ceiling level until the last 10 items of the test, bilinguals showed greater naming variance beginning with item 34, and naming pattern was more erratic. The BNT items also may be inappropriately rank-ordered for this bilingual group, which could be another factor in negative performance outcomes if the test is discontinued prematurely. Quantitative analysis of item difficulty was not possible in the current study due to small sample size but warrants future investigation.
Those raised speaking more than one language seem to be at a disadvantage when it comes to picture naming tasks. Although AoA may contribute to this disadvantage, word frequency, as defined by the frequency with which a person uses or is exposed to a word, likely plays a more salient role. In the current study, more than half of the 19 U.S. born bilinguals acquired English and Spanish simultaneously from birth, and all learned English by age 6 years. Despite early acquisition, language exposure and use over time differs between monolinguals and bilinguals. Whereas monolinguals spend their lifetime utilizing and reinforcing one language, bilinguals practice more than one language. If increased word frequency facilitates naming retrieval, then bilinguals would be disadvantaged, because word use and exposure in this population is distributed across languages. The fact that bilingual adults demonstrate weaker verbal skills in each language compared with monolingual speakers of those languages (Bialystok et al., 2012) highlights the effects of bilingualism on naming skill. In addition, research suggests that the nature of language processing for bilinguals is more effortful than for monolinguals. According to Bialystok and colleagues, fluent bilinguals show a degree of activation and interaction of both languages at all times, even in contexts driven by only one language. This dual activation creates a more complex selection and attentional process at the level of language system that does not exist for monolinguals, making even the simple task of retrieving a common word more difficult for bilinguals (for an in depth discussion, see Bialystok et al., 2012).
The current study has several limitations. First, the BNT was administered in a nonstandard format, such that participants were exposed to BNT objects prior to naming them, and the naming condition was in a written rather than oral format. For the purpose of our study, an unbiased rating of object familiarity was imperative, and so the familiarity component of the study preceded the naming condition without counterbalance. This sequence exposed all participants to the full series of BNT objects without indication that they would later be asked to name them. Although a language-use questionnaire followed, which required a shift in attention away from the BNT objects and created a temporal delay between the familiarity and naming components of the study, prior exposure to the objects may have differentially primed participants in the two groups for the subsequent naming condition. Still, research indicates that bilinguals should have benefited from such priming more than monolinguals. Gollan and colleagues (2005) compared performance on a picture naming task across five repeated trials and found that the drop in error rates for bilinguals from the first to second exposure was more pronounced than for monolinguals. Without the familiarity component in the current study, a stronger naming discrepancy between bilinguals and monolinguals might be expected.
Administration of the BNT also deviated from standard format in that participants were asked to write the names of objects rather than to speak them. This written format uses a different set of cognitive processes and may have affected performance results. However, bilinguals underperform on the BNT compared with monolinguals even when the BNT is administered in oral format (Gollan et al., 2007; Kohnert et al., 1998; Roberts et al., 2002). The current study did not aim to replicate these established findings, rather to explore the degree to which culture and familiarity might explain them. Validating the findings in our study using standard administration procedures may be useful for clinicians interpreting BNT performance results for their bilingual patients.
Our cohort is not representative of the Spanish–English population in the United States, given their high level of acculturation and strong educational background. The current findings, therefore, cannot be generalized to all bilingual groups; however, it is reasonable to expect that bilinguals with less acculturation and fewer years of education will perform worse on the BNT compared to the current bilingual cohort, which further complicates the interpretative process of picture naming tasks in a clinical setting. Given that our sample size for the U.S. born bilinguals is small, future testing of a larger number of individuals is warranted. Future studies might also incorporate an objective measure of language proficiency. The current study measured language proficiency using subjective scales, which carry reporter's bias. Given that word frequency and language proficiency are intimately related, an objective measure of language proficiency will better elucidate how these variables influence BNT performance among bilingual participants.
In summary, the BNT is a widely used instrument in clinical settings and extremely valuable in assessing cognitive–linguistic dysfunction; however, weak performance among bilinguals is difficult to interpret and often does not reflect an acquired deficit. As the BNT is language specific, it may be unsuitable for bilingual populations. Developing different norms and/or reordering the test items may not be possible, given that bilinguals are a heterogeneous group; no single set of norms or single order of item presentation would reflect the entire group's naming ability. Although the current study focuses on the BNT and Spanish–English bilinguals, similar patterns might emerge using other picture naming tasks and with other bilingual or multilingual groups. The current findings underscore the importance of evaluating bilinguals under bilingual classification, using the necessary caution in administering the BNT and interpreting test results. Testing the limits by administering all 60 items to this population is warranted but not enough. Strong performance on the BNT for a bilingual person can be used to exclude certain acquired deficits, but weak performance must be interpreted cautiously, or possibly even discounted in the assessment of an individual patient.
Miami's diverse Hispanic community offers an excellent opportunity in advancing the research on bilingual picture naming performance. Future studies might explore whether and under what circumstances bilingual naming scores improve if given the opportunity to name objects in combined languages. Researchers might also consider investigating whether object naming improves for items with cognate names (translation equivalents, such as canoe in English and canoa in Spanish) and whether a naming test based on cognate names alone might serve the Spanish population better (see Gollan et al., 2007). A picture naming task based on high-translatable words is yet another avenue to explore. Research suggests that bilinguals name high-translatable words (words that bilinguals readily know in both languages) quicker and with fewer errors compared with low-translatable words (Gollan et al., 2005), and this finding may be related to the dual language activation described earlier. While dual language activation is assumed to slow lexical retrieval for bilinguals, it may ultimately provide the key to developing a picture naming task suitable for assessing bilinguals that share languages, but perhaps not cultures (e.g., Spanish–English bilinguals from various ethnic backgrounds).
Conflict of Interest