Sprinting in Two Languages: Picture naming performance of older Spanish-English bilinguals on the MINT Sprint 2.0

Abstract

Objective:

The present study examined how years of immersion in a nondominant language affect a) degree of bilingualism as measured by picture naming scores and b) the bilingual disadvantage relative to monolinguals.

Methods:

Forty-two older Spanish-English bilinguals named pictures in an expanded rapid administration version of the Multilingual Naming Test (MINT Sprint 2.0) in both languages and completed a language history questionnaire. English speaking monolinguals (n=138; from Gollan, Garcia et al., 2023) named pictures in just one language.

Results:

Spanish-dominant bilinguals named more pictures in the nondominant language, but fewer pictures in the dominant language relative to English-dominant bilinguals. Increased years of immersion in the nondominant language increased naming scores in that language but decreased naming scores in the dominant language. When controlling for differences in age and education level, monolinguals named more pictures than bilinguals even in their dominant language, a difference that was numerically smaller for English-dominant bilinguals. However, two bilinguals who stated they prefer to be tested in English scored much higher in Spanish.

Conclusions:

Older bilinguals name fewer pictures than demographically matched monolinguals even when bilinguals are tested in their dominant language and especially if they report many years of immersion in their nondominant language. The bilingual disadvantage can be magnified if self-reported language preference is used to determine the language of testing. Accurate interpretation of bilingual picture naming scores requires a thorough language history, and objective assessment in both languages, which can be done in relatively little time using rapid administration procedures.

A general consensus among neuropsychologists is that cognitive assessment of bilinguals must be conducted in the dominant language to maximize diagnostic sensitivity and minimize false positive diagnoses (Gasquoine & Gonzalez, 2012; Gollan et al., 2010; Gollan, Stasenko, et al., 2023). It is also generally accepted that even when tested in their dominant language, bilinguals may underperform relative to monolingual speakers of the same language, especially on tests that require language production (Gollan et al., 2005, 2011; Palomar-García et al., 2015). However, much of what is known about how bilingualism affects language production has been observed in studies of children (Klassert et al., 2013; Peña et al., 2020) or college students (Gollan et al., 2005; Kohnert et al., 1998), with relatively few studies on older bilinguals who are a growing proportion of those referred for cognitive testing in clinical settings. Precise information about the extent to which bilingualism affects performance of older bilinguals on tests that require language production is lacking, in part because there is no standardized method for assessing bilingualism or even language dominance – particularly for older bilinguals.

While it is well accepted that bilinguals are disadvantaged if they are tested in a nondominant language (Bencivenni et al., 2021; Hanulová et al., 2011; Runnqvist et al., 2011; Sadat et al., 2012, 2015), a recent meta-analysis concluded that bilinguals are disadvantaged only if they are tested in a late-learned second language (Bylund et al., 2023). While meta-analyses can be useful, they often are forced to disregard critically important methodological differences across studies, and often fail to consider many factors that can have substantial effects on performance (such factors are plentiful and varied in bilinguals; Gullifer & Titone, 2020; Navarro-Torres et al., 2021). The Bylund et al. meta-analysis falls into this trap by collapsing studies of picture naming, semantic fluency, and letter fluency. The inclusion of letter fluency is particularly problematic given that previous studies showed that letter fluency is either less affected or not affected by bilingualism relative to semantic fluency (Gollan et al., 2002; Portocarrero et al., 2007; Sandoval et al., 2010), with some studies even showing a bilingual advantage relative to monolinguals in letter fluency (when controlling for vocabulary size, possibly due to enhanced executive control in bilinguals; Luo et al., 2010; Patra et al., 2020). The meta-analysis also did not consider language dominance as a critical factor or other aspects of language history that could determine which bilinguals are more likely to exhibit a disadvantage.

Multiple studies reported a significant bilingual disadvantage in picture naming (Gollan et al., 2005, 2008, 2011), even when bilinguals were dominant in their first learned language (Ivanova & Costa, 2008) or when materials were equally familiar to bilingual and monolingual groups (Misdraji-Hammond et al., 2015). Behavioral and neurological evidence suggests that this disadvantage stems specifically from difficulty with lexical access (Baus et al., 2020; Strijkers et al., 2010). Limitations in previous bilingual picture naming studies include separation of bilinguals by language of testing instead of by which language is dominant (Klassert et al., 2013; Sheppard et al., 2016), and testing in just one language (Sullivan et al., 2018) which leaves an open question as to whether they are disadvantaged in the language that produces their highest naming score (in some cases bilinguals score higher in the language they report as nondominant). Bilingual picture naming might also be affected in these studies by counterbalancing language of testing order (which cannot be done in monolinguals who are tested in just one language; Sheppard et al., 2016). Testing in the nondominant language first could lower scores especially in dominant language (Degani et al., 2020; Misra et al., 2012; Stasenko & Gollan, 2019; Wodniecka et al., 2020, but see Garcia & Gollan, 2022). Finally, many previous studies used the Boston Naming Test (BNT) which was developed for use with monolingual English speakers and is problematic when testing in other languages (e.g., Allegri et al., 1997; Gollan et al., 2012; Kohnert et al., 1998).

A critical next step is to identify factors that can affect naming ability in bilinguals so that interpretation of individual performance can be adjusted accordingly. This is especially important for older bilinguals who often have rich language histories, and for whom there is little or no normative data that adjusts for the many factors that can affect naming performance in bilinguals (e.g., similarity of the languages spoken, age of acquisition, method of exposure, current and past frequency of use, immersion experience, etc.; Bailey et al., 2020). The present study aimed to bridge these critical gaps by examining performance of older Spanish-English bilinguals on the MINT Sprint 2.0 (Gollan, Garcia, et al., 2023), an expanded Multilingual Naming Test with a rapid administration procedure that was developed for use with speakers of English, Spanish, Mandarin, and Hebrew and recently found to be sensitive for detecting preclinical AD in English speaking monolinguals. We tested bilinguals first in their self-reported language of preference for cognitive testing and then in the other language. We then separated bilinguals into Spanish-dominant and English-dominant subgroups based on whichever language produced a higher naming score for each individual. We then compared the two bilingual groups to each other in both languages and to English speaking monolinguals and explored the effects of language history on naming scores while focusing primarily on the dominant language.

Methods

Participants

Forty-one cognitively healthy older Spanish-English bilinguals who were followed by a longitudinal study at the University of UCSD Alzheimer’s Disease Research Center (ADRC) were included (this included all bilingual participants at the ADRC with a Normal diagnosis). Monolinguals included all cognitively normal controls (n=138; Gollan, Garcia et al., 2023) who were tested in our previous study following the same procedures (but named pictures in English only). Diagnoses were made based on annual evaluations with clinical and medical history, brief medical examination, neurological and neuropsychological assessment, screening for depression and other psychiatric symptoms, assessment of functional activities of daily living, and laboratory tests. Results at the ADRC are reviewed by at least two board-certified neurologists and a neuropsychologist to reach a consensus clinical diagnosis. All participants in the present study were classified as cognitively normal via this process (with the exception of one bilingual who had not yet completed the medical and neurological exams, but neuropsychological tests suggested intact cognitive abilities). The MINT Sprint 2.0 scores were not considered in reaching the diagnosis. Table 1 shows participant characteristics.

Table 1.

Demographic variables, cognitive test scores, naming test scores, and language history questionnaire data for bilinguals who scored higher in English (n=16) and for bilinguals who scored higher in Spanish (n=26).

		English-dominant Bilinguals (n=16)		Spanish-dominant Bilinguals (n=26)
		M	SD	M	SD
	Age	75.4	9.4	72.9*	5.7
	Education	16.7	2.4	14.5	3.7
	%female	68.8		61.5
	DRS	139.4	3.3	135.6*	6.3
	MMSE	29.1	1.5	28.7	1.3
	MOCA	25.8	3.1	24.5	2.8
Dominant Language MINT Sprint 2.0	Total score	75.9	3.9	73.2*	3.8
	First pass	71.5	5.4	67.9*	4.0
	Second pass	4.4	2.3	5.3	2.0
	Percent Resolved	60.7	24.1	47.1†	20.1
	First pass Time	2.3	0.6	2.6	0.7
	Total time	3.9	1.3	5.0	1.4
Nondominant Language MINT Sprint 2.0	Total score	46.2	20.1	53.3	12.3
	First pass	42.5	18.2	50.1	11.2
	Second pass	3.9	3.1	3.2	3.0
	Percent Resolved	15.0	13.2	12.3	10.9
	First pass Time	3.0	1.1	3.4	0.9
	Total time	5.7	2.0	6.4	2.7
Years Immersed in	Dominant	69.4	7.7	32.7**	16.0
	Nondominant	4.6	5.7	40.3**	14.1
Age of Regular Use of	Dominant	4.9	5.1	0.6**	1.1
	Nondominant	1.9	5.2	21.5**	10.7
Current Percent Use of	Dominant	85.5	14.0	76.6	18.2
	Nondominant	13.9	13.7	23.4†	18.2
Self-rated Proficiency in	Dominant	6.9	0.3	6.9	0.3
	Nondominant	4.6	1.5	5.3	1.5

Asterisks mark means that are significantly different from the mean shown in the adjacent column on the left

^**p≤.01

^*p≤.05

^†p≤.10

1=almost none; 2=very poor; 3=fair; 4=functional; 5=good; 6=very good; 7=like a native speaker

Consent Statement:

The research protocol was approved by the UCSD Institutional Review Board in accordance with the Helsinki Declaration. Informed consent was obtained at the point of entry into the longitudinal study from all participants.

MINT Sprint 2.0 (as previously described by Gollan, Garcia, et al., 2023)

The MINT Sprint 2.0 test and materials are available at https://osf.io/7r9mq/. The MINT Sprint 2.0 has eight rows of ten color pictures of objects (each approximately 1 to 1.5 square inches) simultaneously presented on a 17×13 inches laminated card (Gollan, Garcia, et al., 2023). The bottom rows contain more difficult items drawn from studies designed to elicit tip-of-the-tongue states (Gollan & Brown, 2006; Stasenko & Gollan, 2019). Like the original MINT tests (Garcia & Gollan, 2022; Gollan et al., 2012), the MINT Sprint 2.0 was designed for speakers of English, Spanish, Mandarin and Hebrew. The MINT Sprint 2.0 contains all the original 68 MINT items with the exception of four items that were replaced for being too difficult (porthole), too easy (hand), or poorly matched across languages (i.e., king and witch are rare in Mandarin). Additionally, mortar and pestle are credited as one point in the MINT Sprint 2.0 if either is produced correctly (instead of requiring both words because the Spanish word for pestle is rare). In the 2.0 version, we also replaced a few of the original MINT Sprint items to avoid pictures with easily guessed cognate names (e.g., gyroscope is giróscopo), which can affect naming in bilinguals (Gollan et al., 2007; Gollan & Acenas, 2004). MINT Sprint 2.0 items are ordered by difficulty, collapsing across languages, based on existing Spanish-English data from Garcia and Gollan (Garcia & Gollan, 2022), and pilot data from approximately ten native Hebrew speakers and ten native Mandarin speakers. Items were swapped locally (moving as little as possible) to avoid having consecutive words beginning with the same sound or rhyming in any of the four languages.

To induce a sense of time pressure, participants are told they have 3 minutes to name as many pictures as they can, as quickly as possible, starting at the top left corner and making their way across each row. If participants take longer than 3–4 seconds on any given picture the examiner says, “keep going” and encourages them to not spend too much time on any one picture. The 3-minute cutoff is not imposed (i.e., participants are given as much time as they need), but most participants require less than 3 minutes to complete their first pass (initial attempt) through the grid. Instructions are:

“I am going to show you 8 rows of pictures. Starting at the top left, try to name each picture from first to last going as quickly as you can without making errors. If you come across one you don’t know-or can’t remember - say “don’t know” and keep going. If the name comes to mind later, you can go back and tell us. You will have 3 minutes¹ to name as many pictures as you can.”

After participants indicate they are finished, they are prompted to try again to name only items they had skipped or named incorrectly during the first pass. This second attempt only at items that were previously missed is called the Second Pass. Instructions are:

“Now let’s see if you can get some of the ones you missed. If you still don’t know, just say ‘don’t know’ and we’ll move on quickly. I’m going to point out some objects that you either skipped or weren’t quite right. Please let me know if the name comes to mind.”

The examiner then points to missed items and asks the participant to try again. No semantic or phonemic cues are provided. If the response was incorrect (e.g., tomato instead of apple) the examiner says, “Take a closer look at this one. Do you have a different name for that?” If a superordinate or subordinate response was provided, the examiner says, “Do you have a more specific name/more general name?” If the participant failed to notice an arrow pointing to a critical part of the picture (e.g., if they said window instead of blinds) the examiner says, “See what the arrow is pointing at here. Do you have a name for that?” Finally, if the participant skipped an item or said, “don’t know” the examiner says, “Did you see this one?” or “Do you know this one?” Time to complete the first pass and the entire test is recorded.

MINT Sprint 2.0 scores include: a) number of correct responses on the first pass, b) number of correct responses on the second pass, c) total correct (the sum of a and b), d) a percent resolved (PR) score which is the number of correct responses on the second pass divided by the number of items missed on the first pass (multiplied by 100). The dominant response, dominant response variants (e.g., plane for airplane), picture-specific variants (e.g., Monarch for butterfly), responses including the target name (e.g., wishing well for well) and regional variants (e.g., torch for flashlight) are counted as correct. Items named correctly on the first pass (even if produced out of order) are credited in the first pass score. Correct names produced only after prompting are credited in the second pass score.

Procedure

Participants completed the MINT Sprint 2.0 as part of the annual ADRC evaluation, which included procedures of the Uniform Data Set (UDS) to collect systematic information from all federally funded ADRCs (Morris et al., 2006; Weintraub et al., 2018). Bilinguals completed the MINT Sprint 2.0 first in the language they stated is their preferred language for cognitive testing followed by testing in the other language. Every bilingual scored at least 6 points higher in one of their two languages. For almost all bilinguals the stated language of preference was also the higher scoring language; however, there were two participants who stated they prefer to be tested in English but obtained higher total naming scores in Spanish (additional details provided below). Unless otherwise stated, these participants were classified as Spanish-dominant in the analyses below. A language history questionnaire was administered the first year of ADRC participation and provided self-reported information about years of immersion, age of acquisition (operationalized as “the age when you first began regularly using the language”), current percent use of each language, and self-rated proficiency level in each language (on a 7 point scale; see Table 1).

Participants were tested individually in a quiet well-lighted room. Audio of the MINT Sprint 2.0 administration was recorded. During testing the examiner recorded pictures named correctly on each pass using a rectangular scoresheet that reproduced the spatial grid and order of pictures on the testing card. The acceptable correct name(s) are printed in the appropriate scoresheet location with room to write any incorrect responses. Response codes were used to record responses (e.g., check mark for correct, a circle for Second Pass prompt needed). This format facilitated scoring in time with the rapid pace of naming. Responses and completion time were checked against the audio-recording by a separate examiner.

Statistical Analyses

Data were analyzed using the Statistical Package for Social Sciences (SPSS) v29². Table 1 shows MINT Sprint 2.0 scores for bilinguals who scored higher in English separately from bilinguals who scored higher in Spanish in each language. Scores shown include number correct on the first and second passes, completion times in minutes, and PR scores which reflect the proportion of failed items on the first pass that were resolved on the Second Pass [(Second Pass score/(80 – first pass score)*100]. We focused our analyses primarily on Total MINT Sprint 2.0 scores (the sum of items named correctly on the first pass and the second pass) which provides the most accurate assessment of the functionally accessible lexicon size in bilinguals (Garcia & Gollan, 2021). Total scores were also most sensitive to preclinical AD in our recent study with monolinguals (Gollan, Garcia, et al., 2023).

We began by examining which MINT Sprint 2.0 scores were sensitive to language dominance effects using paired t-tests within bilinguals. Our first analysis of primary interest compared bilinguals who named more pictures in English than in Spanish (i.e., English-dominant bilinguals; n=16) to bilinguals who named more pictures in Spanish than in English (i.e., Spanish-dominant bilinguals; n=26). This comparison was done with a 2 × 2 ANOVA with Total Correct scores as the dependent measure, language (dominant, nondominant) as a repeated measures factor, and bilingual subgroup (English-dominant, Spanish-dominant) as a between-subjects factor. In this analysis we controlled for education level which was about 2 years higher in English-dominant than in Spanish-dominant bilinguals (see Table 1). For comparison, we then compared the two groups in two similarly structured univariate ANOVAs but with DRS scores as the dependent measure first without then with the education covariate.

Next, we used Pearson bivariate correlations, shown in Table 2, to explore if four key variables of interest from the language history questionnaire were significantly associated with Total Correct scores in each language. We then repeated the 2 × 2 ANOVA comparing English-dominant to Spanish-dominant bilinguals while controlling for years of immersion in the nondominant language.

Table 2.

Pearson bivariate correlations between total correct MINT Sprint 2.0 scores and language history questionnaire variables.

		Total score MINT Sprint 2.0		dominant language				nondominant language
		dominant language	nondominant language	Years of Immersion	Age of Acquisition	Percent Use of	Self-rated proficiency	Years of Immersion	Age of Acquisition	Percent Use of
dominant language	Years of Immersiona	.448**	−.374*	--
	Age of Acquisitiona	.241	−.172	.411**	--
	Percent Use ofa	.218	−.482**	.309	.093	--
	Self-rated proficiencyb	.116	.036	.072	.125	.070	--
nondominant language	Years of Immersiona	−.530**	.412**	−.942**	−.468**	−.361*	−.106	--
	Age of Acquisitionc	.027	.123	−.464**	−.233	−.189	.105	.444**	--
	Percent Use ofa	−.232	.465**	−.325*	−.136	−.996**	−.075	.375*	.208	--
	Self-rated proficiencyb	−.321*	.785**	−.344*	−.108	−.542**	.037	.352*	.068	.545**

^**Correlation is significant at the 0.01 level (2-tailed).

^*Correlation is significant at the 0.05 level (2-tailed).

^a-data only available on n=40 participants

^b=data only available on 41 participants

^c=data available on 36 participants

We then compared bilinguals in their dominant language to monolinguals (n=138 from Gollan, & Garcia, et al., 2023 who were tested following the same procedures as bilinguals at the ADRC). In these univariate ANOVAs we controlled for group differences in age and education level because monolinguals were slightly older (M=78.1, SD=5.5) and more educated (M=17.1, SD=2.2) than bilinguals (see Table 1). We then repeated the comparison of bilinguals to monolinguals on matched subgroups by using the case control matching function in SPSS to select monolinguals from the larger group with exact matching for sex, allowing 3 years of discrepancy in education level, and 5 years for age. This created matches for 38/42 bilinguals who were matched to 38/138 monolinguals in age and education levels (ps≥.20).

Lastly, we subdivided the matched groups and carried our four additional comparisons of a) English-dominant bilinguals to matched monolinguals in the dominant language, b) Spanish-dominant bilinguals versus matched monolinguals in the dominant language, c) prefer-English bilinguals versus matched monolinguals in their stated language of preference, and d) prefer-Spanish bilinguals versus matched monolinguals in their stated language of preference. In all of these comparisons, education did not differ significantly between subgroups (all Fs<1), and age was well matched for comparing English-dominant and prefer English bilinguals to monolinguals (both Fs<1), and monolinguals tended to be slightly but not significantly older (both ps ≥ .14) when comparing Spanish-dominant and prefer Spanish bilinguals to monolinguals.

Transparency and Openness

This study was not preregistered. Data used for analyses are accessible at the following OSF link https://osf.io/z3be4/?view_only=8b5033c392554c76bfa18353108b8ceb. Sample size was determined by the number of cognitively healthy bilingual control participants enrolled and active in the UCSD ADRC longitudinal study at the time of testing. Two participants who likely were Spanish-dominant bilinguals but did not wish to be tested in both languages were excluded.

Results

Table 1 shows performance of the two bilingual groups on the MINT Sprint 2.0. English-dominant and Spanish-dominant bilinguals performed equivalently on brief tests of cognitive status (Mini Mental State Exam or MMSE; Folstein et al., 1975, Montreal Cognitive Assessment or MOCA; Nasreddine et al., 2005), and English-dominant bilinguals scored better than Spanish-dominant bilinguals on the Dementia Rating Scale (DRS; Mattis, 1988, but this difference was not significant when controlling for education level, see below). English-dominant bilinguals also scored better than Spanish-dominant bilinguals on the MINT Sprint 2.0 in the dominant language. The bilingual groups also differed in many language history variables. The English-dominant bilinguals had been immersed in their dominant language for more years and reported currently using their dominant language more often than the Spanish-dominant bilinguals. English-dominant bilinguals acquired their dominant language at a later age, on average, than Spanish-dominant bilinguals, but acquired both languages relatively early in life. Spanish-dominant bilinguals acquired their second language later in life, on average, than English-dominant bilinguals. Both bilingual groups tended to name fewer pictures than their matched monolinguals, but this was especially true for Spanish-dominant bilinguals, the only participant group that was not immersed in their dominant (or only known) language at the time of testing.

Language dominance effects.

Paired t-tests comparing dominant to nondominant scores showed significant language dominance effects with very large effect sizes (Sullivan & Feinn, 2012) for first pass scores, total scores, and PR (Percent Resolved) scores (Cohen’s ds=1.40 to 1.50; all ps<.001). Minutes to complete the first pass, and total completion times also exhibited significant language dominance effects with large effect sizes (Cohen’s d=−.89 and −.70 respectively, ps≤.001). Second pass scores were less sensitive to language dominance effect (i.e., showed less than a medium effect size, Cohen’s d=.40, p=.01).

Comparing MINT Sprint 2.0 Total Scores across Bilingual Subgroups.

Figure 1 shows the mean total correct scores for the analysis comparing English-dominant to Spanish-dominant bilinguals. Bilinguals named more pictures in their dominant than in their nondominant language (F(1,39)=11.83, ${\eta }_p^2$=.23, p=.001), and the two subgroups named similar numbers of pictures overall when collapsed across the two languages (F(1,39)=1.77, ${\eta }_p^2$=.04, p=.19). However, English-dominant bilinguals named more pictures in the dominant language than Spanish-dominant bilinguals but fewer pictures in the nondominant language, a significant interaction between bilingual subgroup and language (F(1,39)=5.27, ${\eta }_p^2$=.12; p=.03). The education covariate was not significant (all ps≥.14 for the main effect and for both interactions). By contrast, English-dominant bilinguals scored higher than Spanish-dominant bilinguals on the DRS (F(1,40=5.08, ${\eta }_p^2$=.11; p=.03), but not after including years of education as a covariate (F(1,39)=2.01; ${\eta }_p^2$=.05; p=.16; the effect of education on DRS scores was significant; F(1,39)=8.68; ${\eta }_p^2$=.18; p=.01).

Figure 1.

English-dominant bilinguals had higher Total MINT Sprint 2.0 scores in the dominant-language but lower scores in the nondominant language relative to Spanish-dominant bilinguals (with dominance determined by whichever language score was higher for each person).

Predicting MINT Sprint 2.0 Total Naming Scores from Language History.

Table 2 shows correlations between four key variables from the language history questionnaire and naming scores in each language. Several language history variables (e.g., years of immersion in the nondominant language, current percent use of the nondominant language, and self-rated proficiency level in the nondominant language) were moderately to strongly correlated with Total MINT Sprint 2.0 naming scores in the nondominant language (all ps<.01). By contrast, only one language history variable was significantly associated with naming scores in the dominant language; bilinguals with more years of immersion in their dominant language had significantly higher MINT Sprint 2.0 naming scores in the dominant language (r=.448). Years of immersion in the nondominant language was also negatively correlated with dominant language naming scores (r=−.530).

Repeating our comparison of English-dominant to Spanish-dominant bilinguals, but focusing exclusively on MINT Sprint 2.0 total scores in the dominant language, English-dominant bilinguals scored higher than Spanish-dominant bilinguals (F(1, 39)=4.19, ${\eta }_p^2$=.10, p=.05; with no effect of education, F<1), but after adding years of immersion in the nondominant language as a covariate, the difference between bilingual subgroups did not approach significance (F<1), and the effect of years of immersion was highly robust (F(1, 36)=7.65, ${\eta }_p^2$=.18, p=.01, note for this analysis the degrees of freedom is smaller because we did not have years of immersion values for 2 bilinguals).

Group differences in dominant language naming scores.

Comparing bilinguals (n=42) to monolinguals (n=138), there was a robust bilingual disadvantage (F(1,179)=32.51, ${\eta }_p^2$=.16, p<.001) with bilinguals naming fewer pictures on average 74.2 (SD=4.0) than monolinguals 77.5 (SD=2.8) and no significant effects of age (p=.11) or education (F<1). The bilingual disadvantage was significant both when including only English-dominant bilinguals, (F(1,150)=5.81, ${\eta }_p^2$=.04, p=.02), and when including only Spanish-dominant bilinguals, (F(1,160)=36.27, ${\eta }_p^2$=.18, p<.001).

Similarly, when comparing the age and education matched groups of bilinguals (n=38) and monolinguals (n=38) there was a robust overall bilingual disadvantage in total correct scores with bilinguals naming 74.2 (SD=4.0) pictures in their dominant language and monolinguals naming 77.3 (SD=2.6) pictures (F(1,72)=14.12, ${\eta }_p^2$=.16, p<.001) on average, with a marginal tendency for higher naming scores in English-dominant bilinguals and their matched monolinguals than in Spanish-dominant bilinguals and their matched monolinguals, (F(1,72)=3.60, ${\eta }_p^2$=.05, p=.06), but no significant interaction between language matching group and participant type, (F(1,72)=1.82, ${\eta }_p^2$=.03, p=.18).

Figure 2 shows the mean total correct scores for four comparisons of interest from within the demographically matched sub-groups of bilinguals and monolinguals. The bilingual disadvantage was not significant when considering only English-dominant bilinguals (n=15; M=75.7, SD=4.0), who named almost 2 fewer pictures than their matched monolinguals (n=15; M=77.6, SD=2.2), F(1,28)=2.52, ${\eta }_p^2$=.08, p=.12).³ However, the bilingual disadvantage was robust when considering only Spanish-dominant bilinguals (n=23; M=73.2, SD=3.7, who named 4 fewer pictures than their matched monolinguals; n=23; M=77.2, SD=2.9, F(1, 44)=16.00, ${\eta }_p^2$=.27, p<.001). Furthermore, when dividing bilinguals by stated language of preference for cognitive testing, both prefer-English (n=17; F(1, 32)=4.43, ${\eta }_p^2$=.12, p=.04) and prefer-Spanish bilinguals (n=21; F(1, 40)=13.47, ${\eta }_p^2$=.25, p=.001) were significantly disadvantaged relative to matched monolinguals. These comparisons required moving two older bilinguals who said they preferred to be tested in English but scored higher in Spanish from the Spanish-dominant group to the “Prefer English” group. In this comparison, the bilingual disadvantage was similarly sized regardless of language preference (3.7 points for those who preferred to be tested in English and 4.0 points for those who preferred to be tested in Spanish).

Figure 2.

Bilinguals obtained lower Total MINT Sprint 2.0 scores in both their dominant (higher-scoring) language and in their stated language of preference for cognitive testing relative to demographically matched monolinguals.

Discussion

The results of the present study demonstrated powerful effects of self-reported years of immersion in the nondominant language on naming scores in both languages. The evidence for immersion effects came in two forms. First, Spanish-dominant bilinguals who were immersed in their nondominant language at the time of testing scored higher in the nondominant language but lower in the dominant language than English-dominant bilinguals who were immersed in their dominant language at the time of testing (a significant interaction between bilingual subgroup and language dominance). Second, years of immersion in the nondominant language was significantly correlated with bilinguals’ naming scores, and the Spanish-dominant bilinguals’ disadvantage in dominant language naming scores relative to English-dominant bilinguals disappeared when controlling for years of immersion in the nondominant language. While years of immersion was associated with naming scores in both languages, other self-reported language history questions were associated with naming scores only in the nondominant language (i.e., percent current use, and self-rated proficiency level). The results of the present study also replicated and extended previous findings that bilinguals name significantly fewer pictures than monolinguals even if bilinguals are tested in their dominant language. In the present study this was operationalized as whichever language allowed each bilingual to obtain a higher naming score (the bilingual disadvantage was larger if bilinguals were tested in their stated language of preference for cognitive testing). Finally, with the exception of second pass scores, all MINT Sprint 2.0 measures exhibited strong language dominance effects with medium to large effect sizes, including first pass scores, total scores, completion times, and percent resolved (PR) scores.

A number of cognitive mechanisms might underlie immersion effects. One possibility is that bilinguals who are currently immersed in their nondominant language are more likely to have spoken in the nondominant language immediately prior to testing which could engage inhibition of the dominant language that persists and influences cognitive testing (Degani et al., 2020; Misra et al., 2012; Stasenko & Gollan, 2019; Wodniecka et al., 2020). Immersed bilinguals might also be generally more likely to inhibit the dominant language to manage competition between languages when immersed in a nondominant language context (Beatty-Martínez et al., 2020; Zhang et al., 2021). A second possibility is that years of immersion could influence naming in both languages through increased cumulative frequency of use of the immersed language and decreased frequency of use of the non-immersed language (Baus et al., 2013, 2020; Gollan et al., 2005, 2008, 2011; Ivanova & Costa, 2008; Runnqvist et al., 2011; Sadat et al., 2012; Strijkers et al., 2010). It is not possible to distinguish which of these alternatives was most influential in the present study, and it is important to note that these explanations are not mutually exclusive.

Figure 3 shows an attenuation of dominant language naming scores with increased years of immersion in the nondominant language that is consistent with both accounts (inhibition and frequency lag). Specifically, both Spanish-dominant and English-dominant bilinguals’ naming scores in the dominant language declined with greater years of immersion. Of note, visual inspection of the two lines appears to suggest weaker decline in dominant language naming scores with each increased year of immersion in Spanish-dominant than in English-dominant bilinguals who had far fewer years of immersion overall. We did not test this interaction because of the relatively small number of participants in the present study, but this would be consistent with both explanations proposed above. That is, frequency of use effects might be expected to decline as lexical accessibility approaches maximum levels (Gollan et al., 2008, 2011), and there might be an upper limit on the amount of inhibition needed to manage competition between languages during extended immersion experience. A similar pattern was observed in a recent study that compared young to older bilinguals where young bilinguals exhibited a stronger effect of each individual year of immersion and had fewer years of immersion relative to older bilinguals (Neveu & Gollan, 2024^a; for additional influence of immersion effects on picture naming scores see Neveu & Gollan, 2024^b).

Figures 3.

English-dominant and Spanish-dominant bilinguals with more self-reported years of immersion in the nondominant language exhibited lower Total MINT Sprint 2.0 scores in the dominant language.

Although we obtained robust effects of immersion experience in both subgroups of bilinguals, Figure 4 suggests that different aspects of language history may have different effects on different bilingual subgroups (again, interpreted with caution given the small number of participants). Specifically, in bilinguals who are already immersed in their nondominant language (i.e., Spanish-dominant bilinguals) there was little effect of self-reported frequency of use of the nondominant language on dominant language MINT Sprint 2.0 total scores, whereas in non-immersed (i.e., English-dominant) bilinguals, those who reported using the nondominant language more often had lower dominant language naming scores. Such a difference might occur if Spanish-dominant bilinguals are forced to use their nondominant language even if they are not very proficient in it (and are engaged in conversations that are not likely to promote increased proficiency level), whereas English-dominant bilinguals only use their nondominant language if they already are more proficient or if regularly exposed to contexts that promote proficiency in the nondominant language. Age of acquisition effects were not significant overall in the present study perhaps in part because all English-dominant bilinguals learned both languages relatively early in life (proficiency in Spanish was determined by other factors). Although later age of acquisition of a nondominant language is usually associated with lower proficiency level, in the present study it was also associated with more years of immersion in the nondominant language (which increased proficiency) which may have attenuated its effect. Finally, while bilinguals’ self-rated proficiency level was significantly correlated with naming scores in the nondominant language, self-ratings were not significantly correlated with dominant language naming scores (for similar findings see Garcia & Gollan, 2022; Gollan et al., 2012; Marian et al., 2007; Neveu & Gollan, 2024^a).

Figure 4.

English-dominant bilinguals with greater self-reported percent use of the nondominant language exhibited lower Total MINT Sprint 2.0 scores in the dominant language. Spanish-dominant bilinguals (who were immersed in the non-dominant language) exhibited no effect.

We focused most of our analyses on MINT Sprint 2.0 Total correct scores in the present study because these likely best approximate the functionally accessible lexicon in bilinguals, were most sensitive to preclinical AD in a previous study, and most closely resemble scores from other standardized tests of picture naming. However, we observed highly robust language dominance effects on several other MINT Sprint 2.0 measures, including test completion times, for both English-dominant and Spanish-dominant bilinguals. Of note, PR (percent resolved) scores also revealed significant and robust language dominance effects such that bilinguals were more likely to resolve items (i.e., correctly name items missed on the first pass when prompted with “try again” on the second pass) in the dominant than in the nondominant language. This is consistent with our previous observation that PR scores are higher in cognitively healthy controls than in participants with preclinical AD or dementia, and with our conclusion that higher PR scores are consistently associated with better overall lexical retrieval ability.

In our previous study (Gollan, Garcia et al., 2023) we found a significant positive correlation between PR scores and older age, which could be mistaken as evidence of lexical retrieval deficits in older age. However, PR scores increased with age only in True Controls (i.e., only in controls who were cognitively healthy and biomarker negative, whereas patients or even just controls who were at risk for developing Alzheimer’s disease based on CSF biomarkers did not exhibit increased PR scores with increasing age). We suggested this implies that lexical knowledge continues to accumulate into older age but only in cognitively healthy older adults. Our finding in the present study of higher PR scores in the dominant language provides further evidence for this interpretation; in between-participant comparisons, higher PR scores signify improved lexical retrieval ability (or at very least do not provide evidence for impaired or disadvantaged lexical retrieval). Note of course that given two people with identical total correct scores, if one has a higher PR score this person would have reduced lexical retrieval ability relative to the person with a lower PR score. However, in the present study, first pass scores were strongly and positively correlated with PR scores (r=.518, p<.001), which means that in general higher PR scores were associated with better (not worse) lexical retrieval.

A common practice in clinical settings is to simply ask bilinguals which language they prefer for testing. In the present study, the answer to this question accurately reflected which language was also objectively dominant in most cases. However, the two exceptions in the present study demonstrated the potentially serious consequences of relying exclusively on self-report. Specifically, these participants preferred to complete cognitive testing in English but scored higher in Spanish on the MINT Sprint 2.0 and were not relatively balanced bilinguals. One scored higher in Spanish by 11 points and the other by 17 points.⁴ These differences are orders of magnitude larger than the bilingual disadvantage in dominant language naming scores, and if misinterpreted, could easily be mistaken for a cognitive deficit instead of a normal effect of language dominance, and could inflate the bilingual disadvantage substantially (e.g., relative to bilinguals who actually did name more pictures correctly in English than in Spanish; see Figure 2). In our previous study, the difference in MINT Sprint 2.0 total correct scores between true controls and those with preclinical AD was only 1.4–1.6 points (see Tables 1 and 5 in (Gollan, Garcia, et al., 2023). The results of the present study, therefore, illustrate the importance of obtaining a thorough language history and testing bilinguals in both languages on at least some objective cognitive measures in order to maximize accuracy of diagnosis in clinical settings and to draw accurate conclusions about how bilingualism affects test performance. This may be particularly true with younger (college-aged) bilinguals who are more likely to overestimate their level of ability in the nondominant language compared to older bilinguals who may have higher standards of excellence when self-rating their proficiency level (see Table 3 in Neveu & Gollan, 2024^a).

Constraints on Generality and Study Limitations

The main limitation of the present study was the relatively small number of participants especially English-dominant bilinguals, and the difficulty of matching to monolinguals in education level (especially the Spanish-dominant bilinguals). We observed numerical tendencies towards a smaller bilingual disadvantage for English-dominant bilinguals who were immersed in their dominant language at the time of testing than for Spanish-dominant bilinguals who were immersed in their nondominant language at the time of testing, but without a significant interaction (additional evidence is needed to test significance with better powered and better demographically matched samples). Additional work will also be needed to replicate the results with bilinguals of other language combinations, and with different languages of immersion (e.g., with bilinguals immersed in Spanish at the time of testing and Spanish speaking monolinguals). Another limitation was that participants were always tested in the higher scoring language first and in the lower scoring language second (with the exception of two participants who asked to be tested in English but scored higher in Spanish). This testing order is likely most comparable to how monolinguals are tested. In previous studies with young bilinguals, language of testing order did not affect picture naming scores (e.g., see Garcia & Gollan, 2022; though testing in the nondominant language did lower verbal fluency scores in the dominant language; Van Assche et al., 2013). Additional work is needed to test if older bilinguals might exhibit stronger order effects in tests of picture naming, however, testing in the dominant language first is likely best to minimize interference effects and maximize diagnostic sensitivity (Gollan et al., 2010; Gollan, Stasenko, & Salmon, 2023; Ivanova et al., 2023).

Conclusions

The results of the present study illustrate the importance of obtaining thorough language history and testing in both languages to maximize interpretability of naming scores in older bilinguals who may have a lifetime of diverse experiences in language exposure and use. Given the possibility of administering the MINT Sprint 2.0 in a relatively short amount of time, it no longer seems justifiable to test bilinguals’ naming ability exclusively in one language during a cognitive evaluation.

Key Points.

• Question: Bilinguals name fewer pictures than monolinguals, but it is not known which bilinguals are more likely to exhibit this disadvantage, especially in older bilinguals who are relatively understudied.

• Findings: English-dominant bilinguals who were immersed in English at the time of testing named fewer pictures in their nondominant language, and more pictures in their dominant language relative to Spanish-dominant bilinguals, but the difference between groups disappeared when controlling for lifetime immersion experience.

• Importance: Bilingualism presents the cognitive system with a measurable challenge that is stronger for bilinguals with more years of immersion in their nondominant language, which raises naming scores in the nondominant language but lowers naming scores in the dominant language.

• Next steps: While it is not possible to develop separate normative data with consideration of all the many factors that affect bilingual naming scores, additional information is needed about which aspects of language history affect different types of bilinguals the most, and about how to adjust expectations accordingly in neuropsychological evaluations.

Data are publicly available:

https://osf.io/z3be4/