Bilingualism is a gradient of experiences that show significant variation across individuals who speak more than one language (DeLuca et al., 2019). This inter-individual variation is evident along several axes between first- (L1) and second-acquired (L2) languages, including proficiency and daily usage, especially when considering unbalanced bilinguals. As the incidence of acquired brain injury (ABI), e.g., stroke, increases (Katan and Luft, 2018) leading to language impairment in aging bilingual populations, it can be expected that bilingual people with aphasia (BPWA) will comprise a greater share of caseloads in forthcoming years (Centeno et al., 2020).

Examining the influence of bilingual language experience on the performance of BPWA on linguistic and cognitive standardized assessments and experimental tasks remains an important pursuit to address basic research questions about language in the bilingual brain post-ABI, and to personalize care and predict individualized outcomes. However, understanding and quantifying language experience in bilingual individuals remains a highly complex undertaking (see Silva-Corvalán and Treffers-Daller, 2015; Köpke and Genevska-Hanke, 2018 for a review), a factor that is further complicated in individuals with ABI for whom language experiences may change following aphasia onset (Peñaloza et al., 2019). Several resources and assessments are often employed to elaborate dimensions of language experience (e.g., use, confidence, and proficiency) and address important questions in bilingual research (Francis, 2021). However, there is no clear consensus in the field on how to survey and quantify language experience in healthy bilinguals (HB) and BPWA, suggesting a need for consistency when collecting and reporting these measures across different studies (e.g., Kašćelan et al., 2022).

Finally, large-scale recruitment of BPWA is a challenging undertaking given the resources required to process participants through studies. As Kašćelan et al. (2022) suggest, bilingualism is an intricate construct and therefore variability among different approaches has arisen in operationalizing bilingualism for research purposes. Accordingly, transparency in methods and conceptualization of measures is relatively low, resulting in controversies such as the equivocal presence of bilingual advantage in executive function (Marian and Hayakawa, 2021). While various datasets exist relating to aphasia (e.g., Mirman et al., 2010; MacWhinney et al., 2011) and multilingualism (e.g., de Bruin et al., 2017) separately, few large datasets exist representative of both populations simultaneously. Furthermore, much of the presently available bilingual data lacks more detailed information. For example, Surrain and Luk (2019) examined 186 studies published between 2005 and 2015 to examine the features of bilingual experience reported. While 79% of studies reported general information about language use at home, a minority (39%) reported this using proportions. de Bruin (2019), reviewing different language history questionnaires and measurement tools, also found that for metrics like age of acquisition (AoA), proficiency, and language use, there was a significant lack of definitional congruency, e.g., “late” vs. “early” bilingualism.

To that end, the aim of this Data Report is to introduce LEX-BADAT: Language EXperience in Bilinguals with and without Aphasia DATaset. We used the language use questionnaire (LUQ; Kastenbaum et al., 2019)—which uses continuous scales and has shown predictive value regarding lexical access in both HB (Kastenbaum et al., 2019) and BPWA (Peñaloza et al., 2019)—and explicitly defined its metrics. We provide summaries of the two datasets included, one of 85 BPWA and one of 31 HB. Additionally, given that the data are multidimensional, we present the results of principal component analyses (PCA) on raw LUQ data for ease of use in statistical analyses (e.g., generating component scores for participants not included in the analyses can be accomplished using the provided loading matrices), and to detect latent variables summarizing bilingual language experience in these two groups. As a result, this is both (i) the largest dataset of language experience in Spanish-English BPWA, and (ii) the largest dataset that includes both HB and BPWA using a shared instrument with directly comparable scales. We believe these data meaningfully inform researchers about the structure of language experience in linguistically heterogenous populations primarily in the United States, and perhaps more importantly, on the underexplored language experience of bilinguals whose access to language is impaired.

The LEX-BADAT dataset comprises a comprehensive collection of bilingual language experience metrics in Spanish-English HB and BPWA using the LUQ (Kastenbaum et al., 2019). Four PCAs were conducted on LUQ self-reported measures to (i) parsimoniously reduce the dimensionality of the raw LUQ data for ease in statistical analyses using this dataset, and (ii) identify the latent variables underlying the assessment of Spanish-English bilingual language experience. For BPWA, results of the PCA for both L1 and L2 revealed two components in each language with eigenvalues >1.0. For HB, results of the PCA for L1 revealed one component with an eigenvalue >1.0, and the PCA for L2 revealed two components with eigenvalues >1.0.

Given the variability across studies in quantifying and referring to different language experience metrics, the components were interpreted in terms of the information they were likely conveying. Thus, components containing any combination of the following metrics: AoA, Educational History, Family Proficiency, and Lifetime Exposure were labeled as Background. The label Use was given to components containing Daily Use (pre- and/or post-ABI Daily Use for BPWA). The label Ability was given to components containing Language Ability Ratings (LARs; pre- and/or post for BPWA). The label Confidence was given to components containing Lifetime Confidence, and labels Family Proficiency or Exposure were given to components that contained Family Proficiency or Lifetime Exposure independent of other Background metrics.

Using the aforementioned labels, the PCA for L1 in BPWA could be separated into two groupings: Use/Background and Confidence/Family Proficiency. Here, L1 Use/Background consisted of pre- and post-ABI Daily Use, Lifetime Exposure, and Educational History. L1 Family Proficiency/Confidence comprised Family Proficiency and Lifetime Confidence. The PCA for L2 in BPWA could be divided into L2 Background/Confidence and L2 Ability/Use. L2 Background/Confidence consisted of L2 AoA, Family Proficiency, Lifetime Exposure, Educational History, and Lifetime Confidence, while L2 Ability/Use was comprised of pre- and post-ABI LAR and pre-ABI Daily Use.

For HB, the PCA for L1 consisted of one component called L1 Ability/Background/Confidence. This component was comprised of LAR, Educational History, Lifetime Exposure, and Lifetime Confidence. The PCA for L2 revealed two components: Background/Confidence and Ability/Use/Exposure. L2 Background/Confidence was comprised of L2 AoA, Family Proficiency, Educational History, and Lifetime Confidence. Finally, L2 Ability/Use/Exposure consisted of LAR, Daily Use, and Lifetime Exposure.

Multivariate techniques, such as common factor analysis and PCA, are commonly applied in aphasia research given the inherent multifactoriality of symptoms secondary to neurological impairment (Gilmore et al., 2019; Meier et al., 2019; Wilson and Hula, 2019).

Previously, Peñaloza et al. (2019) used component scores from a PCA of L1 and L2 LUQ metrics of 27 Spanish-English BPWA (included in LEX-BADAT) to examine the relationship between pre-stroke language proficiency and post-stroke lexical-semantic performance. The authors found that the component scores extracted were significantly predictive of post-stroke lexical-semantic performance in both L1 and L2, both in simple regressions and as an interaction term with the language (i.e., L1 or L2). Further, they revealed that L1 pre-stroke proficiency (in component scores) was a better predictor for BPWA whose L1 was English, whereas L2 pre-stroke proficiency was a better predictor of L2 lexical-semantic performance for BPWA whose L1 was Spanish, possibly reflective of the heterogeneity in language experience and performance secondary to language immersion in the United States among the population.

Furthermore, using similar methods as those described here, Carpenter et al. (2021) found that BPWA language experience in L1 and L2, including Background, Use, and Exposure in L1, and Background, Use, Environment, and Exposure in L2, significantly predicted semantic switching performance during verbal fluency tasks, implicating the relationship between pre-ABI language experience and post-ABI language and semantic executive control.

The potential and utility of LEX-BADAT lies on its relatively large sample size including both BPWA and HB across varying profiles of bilingualism and the resulting scores derived from the PCA conducted independently in their L1 and L2. Performing dimensionality reduction is problematic for studies with smaller sample sizes; however, LEX-BADAT may offer a more reliable solution to address important research questions with BPWA using a manageable number of variables representative of their history of bilingualism. For example, Falconer et al. (in preparation), investigated the relationship between language experience measures and translation performance in BPWA using the component scores presented here. Additional component scores for participants not included in the present analyses, obtained by projecting normalized LUQ data onto the previously established PCA space, were found to predict translation performance. Thus, the components reported here can serve as standardized metrics of language experience that provide the opportunity to generate novel predictors, e.g., component scores corresponding to our principal components for participants not originally included in the analyses.

LEX-BADAT consists of individual-level data from a large group of BPWA and a group of healthy bilingual speakers who completed the LUQ (Kastenbaum et al., 2019) to assess their bilingual language experience. This dataset also provides component scores resulting from a PCA conducted on their L1 and L2 which can be used to inform analyses seeking to examine the influences of language experience on performance on, for example, lexical-semantic tasks (Peñaloza et al., 2019), and lexical access and cognitive control via verbal fluency tasks (Carpenter et al., 2021). We hope that in transparently describing our resources and methodology, we improve practices in the quantification of bilingual language experience and subsequent use of terminology (e.g., Background) when referring to predictors in statistical analyses employing this set of data.

Finally, similar to other datasets relating to language experience, e.g., BEST (de Bruin et al., 2017), LEX-BADAT provides broad insight into the structure of Spanish-English language experiences of bilinguals in relatively linguistically diverse, primarily American communities, and specifically into the experiences of post-ABI bilinguals experiencing aphasia. In considering post-ABI language experience, our analyses are a holistic view of a critically underserved clinical population, capturing linguistic activity both before and after the onset of a period of chronically disordered lexical access.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

The studies involving human participants were reviewed and approved by Ethical Committee of the University of Texas at Austin and the Ethical Committee of Boston University. The patients/participants provided their written informed consent to participate in this study.

MM, EC, and SK contributed to conception of the project and methods of pre- and post-processing of data. MM and EC curated the data and wrote the first draft of the manuscript. MM carried out the statistical approach. MM, EC, IF, and CP contributed to writing the manuscript. All authors contributed toward acquisition of data. All authors contributed to manuscript revision, read, and approved the submitted version.

CP was supported by the Juan de la Cierva-Incorporación Program (IJC2018-037818) funded by Ministerio de Ciencia e Innovación, Agencia Estatal de Investigación MCIN/AEI 10.13039/501100011033. This work was supported by the National Institutes of Health/National Institute on Deafness and Other Communication Disorders through grant T32DC013017 awarded to MM and grant U01DC014922 awarded to SK. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

SK serves as a consultant for Constant Therapy Health with no scientific overlap with the present study. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.