Reframing Bilingual Acquisition and Theory: An Insider Perspective Through a Translanguaging Lens

Abstract

Purpose:

This exploratory study developed a process for reinterpreting previously published research studies in the bilingual literature. Three previously published studies on bilingual phonological acquisition were revisited due to the following characteristics: (a) they applied a theoretical framework for bilingual speech production developed by white bilingual researchers, the dual-systems hypothesis, and (b) project data were interpreted without the input and perspective of researchers representative of the community being studied. This study aims to provide a guide for the readership to reinterpret developmental speech and language studies on bilingual children through (a) the theoretical framework of translanguaging, which was developed by minoritized bilingual scholars and members of the community being studied, and (b) community Insider lenses, or the perspectives of research team members whose lived linguistic experiences match those of the target population studied.

Method:

Original interpretations of data were reexamined and reinterpreted incorporating (a) a research team member from the target community and (b) a novel theoretical lens developed by members of the target community called translanguaging.

Results:

Original findings were extended through the application of translanguaging as a theoretical lens. New interpretations of original data were uncovered when a researcher from the Latinx community was involved in the data interpretation process. New insights were gained on phonological acquisition in bilingual Spanish–English–speaking preschoolers by applying a reinterpretation framework.

Conclusions:

Differences in data interpretation reveal that translanguaging may improve understanding of languaging in bilingual/multilingual communities. Implications for development of representative research teams when examining minoritized pediatric populations are also discussed.

Research in bilingual phonological acquisition is an area of speech, language, and hearing sciences that is understudied as compared to published research examining phonological acquisition in monolingual, English-speaking populations. According to Kohnert and Medina (2009), only 12 studies included bilingual children with speech sound disorders in their study samples at the time their article was published. Over the past 13 years, that number has shown only modest growth. Comparatively, roughly 150 articles examining monolingual language acquisition are published every year. There is clearly a bias in who is performing research and what populations they are choosing to study. Due to the lack of evidence-based assessment methods and testing protocols for bilingual children, clinicians rely on research articles that report normative information on bilingual speech and language acquisition for clinical decision making. The main goal of this review article is to guide clinicians, as critical consumers of research, in the evaluation of existing literature on phonological development in bilingual children and to view bilingual research through a community-based lens.

Research in Bilingual Phonological Acquisition

In the 2020 American Speech-Language-Hearing Association (ASHA, 2021) Member and Affiliate Profile, 91.6% of the reported ASHA members in speech-language pathology identified as white, whereas the remaining 8.4% identified as Black, Indigenous, Asian, Pacific Islander, or multiracial. Looking specifically at Latinx identity, only 6.2% of ASHA speech-language pathology members identified as Latinx. Further, only 0.4% of ASHA speech-language pathologists (SLPs) identified “researcher” as their primary employment. At face value, it could be strongly implied that most researchers are white and, given how little bilingualism is studied in our field compared to English-only speakers, the bilingual research that does exist is often being designed and implemented by white researchers.

This discussion takes into consideration not only the study samples examined in this research but also who exactly is designing, implementing, and interpreting these studies. In other words, are the researchers who study bilingualism bilingual themselves? Does their research team include members from the community or speakers of the languages they study? Insider members of the research team (i.e., individuals who share similar or same communities as the populations being studied) are necessary for culturally responsive research design, data collection, and interpretation (Dillard, 2010; Dwyer & Buckle, 2009; Helme, 2021). Considering ASHA's racial and ethnic demographics, many researchers would be deemed as Outsiders, another term identified from previous literature that is relevant to this study (Dwyer & Buckle, 2009; Helme, 2021). In other words, Outsiders are individuals who may or may not know the language and do not have the lived experiences of those in the communities being studied. Therefore, the presence of Insiders on a research team may provide insight into what cultural and linguistic practices are shared and embraced in the participants' communities, as well as the impact of structural racism on language learning and use in minoritized communities (e.g., García et al., 2021).

This concept ties well into the concept of positionality where a researcher (in this context) acknowledges and understands their social position relative to communities being studies (Bourke, 2014). The way in which an individual is positioned in society acts as a lens through which we interpret the world including the data analysis that researchers perform (Jacobson & Mustafa, 2019). For example, the first author is a Latinx heritage speaker of Spanish and English, positioning them within the communities discussed in this study, identifying them as an Insider, whereas the second author is racialized white and learned Spanish in an academic setting, positioning them further from the community, identifying her as an Outsider. Moreover, this presence of Insiders falls under the idea curated by disability activists of, “Nothing about us without us” (Charlton, 1998, p. 25). In other words, research with bilingual speakers for bilingual speakers, a concept that will be touched on throughout this study.

The Raciolinguistics of Bilingualism in Schools

Being bilingual in an educational setting involves a variety of labels that attempt to capture language exposure and use. According to the National Center on Education Statistics (2015), 78.3% of bilingual students come from racially minoritized communities. Therefore, race, ethnicity, and bilingualism are inextricably linked. Terms such as English as a second language, English language learner, long-term English language learner, and semilingual have been used in discourse regarding bilingual speech-language pathology as well as the field of education (Przymus, 2022). What many of these labels have in common is that they imply a sense of deficiency in English at their core and suggest the need to change the speaker's languaging practices (i.e., dialogic communication activities that construct meaning; Swain, 2010) to reflect the standard (Flores et al., 2015). This brief definition of raciolinguistic ideologies conceptualizes home language practices of racialized communities as inherently deficient (Flores, 2020). Therefore, many bilingual students experience raciolinguistic ideologies that perpetuate linguistic deficiency, therefore modifying how people language is the solution to racial oppression rather than modifying the system in place that centers whiteness (Rosa & Flores, 2017). In other words, a raciolinguistic ideology enforces a push toward a “monolingual-likeness” in English.

The SLP's Role: Misdiagnosis of Communication Disorders

What is the connection between raciolinguistic ideologies and communication disorders? How does misdiagnosis occur in the clinical context? Misdiagnosis of speech and language disabilities in the Latinx pediatric population was reported in depth by Artiles et al. (2009). Misdiagnosis is not limited to speech and language, but also learning disabilities and other special education categories. They reported that misdiagnosis occurred not only for bilingual Latinx children but also for Black and Indigenous children as well. Artiles et al. (2009) specify that misdiagnosis of speech and language disorders stems primarily from the scores derived from standardized tests normed on and administered only in English. We focus here specifically on speech sound disorders and the motivating factors for misdiagnosis in this particular domain of communication.

What is the clinician's role in misdiagnosis? Clinicians may perceive certain differences in the communication abilities of bilingual children, relative to monolingual English-speaking children, as errors for remediation. Specifically, bilingual children who are evaluated by untrained clinicians may exhibit communication behaviors that are the result of knowing multiple languages and are not evidence of a true communication disorder. Clinicians, however, may deem those same behaviors as clinical red flags (Przymus & Alvarado, 2019). For example, producing an approximant [ɹ] in place of a trilled /r/ is typical in the speech of Spanish–English–speaking bilingual children (e.g., Fabiano-Smith & Barlow, 2010). This substitution, however, may be scored as an articulation error on a single-word standardized test. To place this example in the context of raciolinguistics, clinicians are viewing students as deficient in their speech production when that aspect of speech is common in their linguistic community. Fabiano-Smith et al. (2014) examined dialect use in Puerto Rican Spanish-speaking preschoolers in Northeast Philadelphia and found that children routinely substituted the flap [ɾ] for the trill /r/. This substitution could be perceived by clinicians as a reduction of the trill, avoidance of the most complex phoneme in the Spanish inventory, or the absence of the multiple vibrant in a child's phonemic inventory. Taking into consideration sociolinguistic context, the flap for trill substitution is a common way of producing Spanish in this linguistic community (Sieg et al., 2023). Context-free clinical judgments can contribute to overidentifying (i.e., identifying a child who is typically developing as presenting with a communication disorder) racially minoritized students for special education services (i.e., speech and language services). Overidentification results in educational disparities. This plays out as bilingual children being removed from the classroom to receive unnecessary speech and language intervention services. Lack of access to the general education curriculum, as compared to their white monolingual English-speaking peers who are correctly identified as typical, has long-term consequences. The greatest single predictor of post-school outcomes is access to the general education curriculum (Test et al., 2009). In addition, the intersectionality of race with disability directly contributes to the school-to-prison pipeline (Connor et al., 2016; Stanford & Muhammad, 2017). To summarize the bigger picture, the bias inherent in special education practices stigmatizes bilingual children's languaging abilities and places them at risk for misdiagnosis.

Misdiagnosis is complicated by the fact that it can negatively impact children in different ways. On one hand, there are a disproportionate number of bilingual students in special education due to the common use of standardized tests (normed using monolingual English-speaking children) to diagnose speech and language disorders in bilinguals (e.g., Haynes & Pindzola, 2011). On the other hand, students are also not being assessed due to their bilingual status or are underidentified as presenting with a speech or language disorder. More specifically, clinicians report a lack of bilingual training and therefore avoid assessment of bilingual children (Skahan et al., 2007). In addition, patterns of underidentification have been observed in school districts that report a high number of Latinx students (e.g., Artiles et al., 2005). Monolingual English-speaking, non-Latinx clinicians who employ the use of standardized tests of English in predominantly white school districts are likely to use those same assessment methods with bilingual children, leading to overidentification. Conversely, districts with large numbers of Latinx children may not quickly identify bilingual children with communication disorders because unexpected productions could be explained away as “bilingualism.” García (2009) outwardly states that, “When officials and educators ignore the bilingualism that these students can—and must—develop through schooling in the United States, they perpetuate inequalities in the education of these children” (p. 322). With this in mind, we begin to see that the lack of education on bilingualism (i.e., insufficient research and bureaucratic diagnostic decision making) leads to different actions, all included in the category of misidentification, that in turn perpetuate inequality and increase disparities in the special education system.

Theory and Practice: Defining the Dual-Systems Model

The majority of studies documenting speech sound development and disorders in bilingual children over the past 20 years have applied a common theoretical model: the dual-systems hypothesis (also referred to as the dual-systems model). This framework has appeared in many research articles on bilingual phonological acquisition across a variety of languages (e.g., Paradis, 2001; Paradis & Genesee, 1996). The dual-systems model postulates that, during speech and language development, languages are differentiated and maintained as two separate systems with points of interaction, or what Paradis and Genesee (1996) referred to as interdependence. The interdependence between languages is predicted to occur in the following three ways: phonological transfer, acceleration, and deceleration.

Transfer is an interaction characterized by the transfer of speech sounds between linguistic systems, for example, using the English [ɹ] in place of the Spanish trilled /r/ in “carro” /karo/→[kaɹo] (Fabiano-Smith & Goldstein, 2010b). Acceleration refers to rate of phonological development, suggesting that speech sound development can occur at a faster rate as compared to monolinguals because of the interaction between the two phonological systems aiding the acquisition of the child's languages (Fabiano-Smith & Goldstein, 2010b; Paradis & Genesee, 1996). The third category, deceleration, is a foil to acceleration; it is defined as the slower rate of phonological development in bilinguals as compared to monolinguals (e.g., Fabiano-Smith & Goldstein, 2010b). Paradis and Genesee (1996) posit that deceleration occurs due to interference, or conflicting rules, between phonologies during production.

It is important to clarify that deceleration does not refer to a clinical property of bilingualism; more specifically, while bilingual children have been found to demonstrate a slower rate of acquisition on some measures, at certain points in acquisition, bilingual children do not demonstrate phonological skills at the level of clinical concern (i.e., in the disordered range). Skills are still considered to be in the typical range for their chronological age, in both languages, but when each language is compared separately to monolinguals, their rate of acquisition appears slower. In the studies of Fabiano-Smith and Barlow (2010) and Fabiano-Smith and Goldstein (2010b), researchers found that the English phonetic inventories of Spanish–English bilingual 3-year-olds were acquired within expected timeframes when compared to monolingual English-speaking children but showed overall lower accuracy in production of consonants as compared to monolinguals. The same result was found for the Spanish phonetic inventories of the bilingual children when they were compared to monolingual Spanish-speaking children of the same age on measures of consonant accuracy.

In the context of phonological assessment, this framework provides the rationale for performing a contrastive analysis in which points of interaction (e.g., transfer) are not scored as errors but as between-language interaction (Goldstein & Iglesias, 1996). Assessments like the Bilingual English–Spanish Assessment (BESA; Peña et al., 2018) utilize this framework to allow for single-word tests to be scored in tandem with expected cross-linguistic influences. Furthermore, treatment design for speech sound disorders in bilingual children may apply this theoretical approach (Kohnert & Goldstein, 2005). Treatment goals can be developed separately for each language, and then, phonological elements common to both languages can be targeted with the hope of facilitating generalization from one language to the other. Overall, the dual-systems model posits the idea of separated systems with points of interaction, and as an outcome, clinicians are encouraged to evaluate each system separately while considering how and when they may interact.

Defining Translanguaging

Recent research explores a novel way of framing bilingualism that might more accurately reflect how bilingual children use language. Translanguaging is defined as the deployment of an individual's linguistic repertoire, or idiolect, in communication (Otheguy et al., 2018). The “trans-” in translanguaging reflects this notion of moving beyond language (Otheguy et al., 2015). This framework, originating from the field of education, negates the bounded notions of named languages allowing for individuals to deploy linguistic features without closely regulating the social and political notions curated by named languages (Otheguy et al., 2018). In this study, translanguaging is a lens through which previous work on bilingual phonological acquisition was reexamined. The authors purposefully selected “lens” for the scope of this review article to reflect the use of the educational pedagogy as a looking glass through which we reinterpret previous findings.

Named languages, often referred to as in translanguaging literature, are language labels used to distinguish one grammar, or code, from another. For example, some named languages include “Spanish,” “Portuguese,” and “Swahili.” From a linguistic anthropological perspective, named languages are human made; anthropologists, linguists, and other individuals created boundaries between languages and categorized them together. In other words, they observed individuals speaking, or signing, to each other in a code and created a delineated category to bind that code as “Language X.” With that, language is something that can be quantified and separated, as is posited in the dual-systems model. However, translanguaging takes a different perspective, suggesting that individuals have a single unitary system that composes the linguistic repertoire (Otheguy et al., 2018). Moreover, when suggesting a unitary system, the framework is a system that is unified and exists without boundaries; linguistic features exist in a pool, not separated by the bounded notions of named languages. For the purposes of defining translanguaging, a unitary model describes the unity of linguistic features, unbounded from one another, in a single repertoire. Not to be confused with the unitary-system model of phonological representation (e.g., Vihman, 2002), translanguaging does not posit one unitary system that eventually differentiates into two separate language systems by the end of the preschool years. On the contrary, translanguaging posits an idiolect that is developed over time as a single repertoire of features, some that overlap with some interlocutors but perhaps not others, that is specific to an individual, unique speaker. The idiolect, as Otheguy et al. discuss, is an internal perspective to language, which individuals separate from external perspectives of languages. This perspective is based, in part, on the idea that the separation of named languages is not a natural process for humans (Otheguy et al., 2018). External perspectives of language are tied to the social factors that categorize and define named languages (Otheguy et al., 2015).

The main component of translanguaging is the concept of an idiolect. An idiolect is an individual's own unique language composed of components, like phonology, and subcomponents, or systems of tenses, for example (Otheguy et al., 2015). Though not addressed within the scope of this review article, interlocutors using non-speaking modalities (e.g., signed languages) also have their own idiolect composed of different components that make up a linguistic repertoire such as that of spoken modalities. Additionally, Otheguy et al. add that the idiolect develops, and deploys, because of social and personal interactions. The deployment of a speaker's own linguistic repertoire is a social behavior in that a speaker will deploy linguistic features that overlap with the repertoire of the other interlocutor; therefore, language use cannot be observed outside of its social context. The majority of studies examining bilingual phonological acquisition have examined named languages as separate entities, have aggregated speech samples from multiple Spanish-speaking communities, and quantified production measures, treating phonology as an extractable (from social context) phenomena (e.g., Fabiano-Smith & Barlow, 2010; Fabiano-Smith & Goldstein, 2010a, 2010b). Additionally, the comparison in previous studies between monolingual and bilingual speaker groups conflicts with the idea of the idiolect. Comparing an individual's languaging practices to that of their monolingual counterparts does not fit within a translanguaging lens because bi- and mono- categories do not exist—only unique idiolects that house a set of features. This novel lens puts into question the use of monolingual groups as comparison groups for bilinguals in research.

Translanguaging considers the phonological repertoires of all language skills without ranking them against each other in terms of strength or weakness, perpetuating the linguistic discrimination that occurs with bilingual speakers. This rank order is often an outcome of how the dual-systems model is interpreted by researchers. Recall that 78.3% of bilingual children are from racially and ethnically minoritized communities. Translanguaging works in tandem with the raciolinguistic perspectives to allow racially minoritized students to engage in their own individual languaging practice without feeling policed by forces like “Standard American English” or “Academic Language.” In line is the perspective mentioned in the work of Flores (2020) about students as “language architects.” Students know and understand their own languages and can manipulate and make decisions to effectively communicate based on interlocutor. Translanguaging allows bilingual children liberation from standards and moves toward framing bilinguals as knowledgeable in their forms of communication. This ideology is the lens through which previous studies are reinterpreted with the hope of extrapolating deeper knowledge of phonological acquisition in this population.

Purpose

The purpose of this study was to reexamine previously published studies examining bilingual phonological acquisition. The main goal of this study was to provide SLPs with a critical guide for evaluating published research on bilingual phonological acquisition and disorders, specifically those studies published by all-white research teams. The studies selected were interpreted using the dual-systems model as the main theoretical framework and conducted by all-white, non-Latinx research teams. This study reinterpreted stated findings of selected studies through the lens of translanguaging. The research questions that framed this study included the following.

1. What linguistic phenomena can be uncovered when shifting data interpretation from a dual-systems model framework to a translanguaging lens?

2. Does the community-based wisdom of an Insider researcher uncover linguistic phenomena in previously reported findings by all-white research teams?

This study hypothesized that the translanguaging lens would provide novel interpretations of previously reported findings. These new interpretations would provide a deeper understanding as to how bilingual phonology is acquired because the phonological organization and use of bilingual speech under translanguaging reflects the naturalistic way that bilingual communities use speech in context. Additionally, this study hypothesized that Insider researchers aiding in the interpretation of data could reveal additional phenomena that were missed by Outsider researchers. Without the personal experience of living and communicating in bilingual communities, phenomena that are easily identified by members within those linguistic communities may be unidentifiable to Outsiders. Clinically, our goal was to expand upon our theoretical rationale for bilingual service delivery and to reduce misdiagnosis and disproportionality in bilingual children attending U.S. schools.

Method

Article Selection and Reinterpretation Process

The purpose of this study was to reinterpret a selection of previously published research, performed by the second author and her colleagues, in bilingual phonological acquisition. Previous studies were reinterpreted in terms of (a) theoretical framework and (b) Insider lens.

Article Selection Criteria

Research articles for review were selected using the following criteria: (a) the topic area of bilingual phonology examined using quantitative phonological measures of speech sound production commonly used by clinicians; (b) the dual-systems/interactional dual-systems model as a theoretical framework for investigation and interpretation of findings; and (c) research studies conducted by all-white, non-Latinx research teams in bilingual Latinx communities. The selected studies are not an exhaustive list of research on this topic but rather studies that have originated in the Fabiano-Smith research laboratory over the past 12 years that employed the dual-systems model and supporting literature using the same theoretical framework (Paradis, 2001). As a result, the following three articles were selected: Fabiano-Smith and Goldstein (2010b), Fabiano-Smith and Barlow (2010), and Fabiano-Smith and Bunta (2012). All three of the studies investigated Spanish and English bilingual phonological acquisition using the same sample of bilingual Latinx Spanish–English–speaking children from Philadelphia, Pennsylvania. None of these three studies included a research team member from the community being studied, and none of the researchers spoke Spanish as a heritage language. All research team members across the three studies identify as white and non-Hispanic/Latinx.

Review of Selected Literature

Fabiano-Smith and Goldstein (2010b)

Fabiano-Smith and Goldstein (2010b) investigated the interaction of phonological systems in preschool-age Latinx bilingual Spanish–English–speaking children. This study used the dual-systems model as a theoretical framework and cited evidence of between-language interaction (also known as the interdependence) in the form of three hypotheses proposed by Paradis and Genesee (1996). As discussed, and defined above, these interactions included the hypotheses of acceleration, deceleration, and transfer. In addition to these types of interactions, Fabiano-Smith and Goldstein (2010b) also discussed variations in acceleration. They argued that there was evidence to suggest that bilingual children demonstrated rates of acceleration like that of the expected ranges for monolingual speakers of both languages. Another point of discussion raised in this study was evidence that all three interactions were observed simultaneously. In combination with a study published by Fabiano-Smith and Barlow (2010), researchers identified that the same set of bilingual children demonstrated lower consonant accuracy for both languages relative to their monolingual counterparts but, at the same point in development, also exhibited an equivalent level of phonetic inventory complexity. This finding suggested that deceleration was occurring in tandem with a variation of acceleration and/or transfer (Fabiano-Smith & Barlow, 2010; Fabiano-Smith & Goldstein, 2010b). The last key finding from this article was the concept of phonetic similarity. The study grouped sounds between Spanish and English as shared (i.e., found across both languages) and unshared (i.e., sounds only found in one of the languages). Much like in the first article discussed in this section, researchers postulate that sounds that are phonetically similar across the two languages will fit into the same category (e.g., Flege, 1995). Under this prediction, phonetically similar sounds were more likely to be employed as they can be quickly accessed but only if the sound did not carry a higher level of difficulty into the production of the other language. Finally, frequency of occurrence of shared sounds was an additional component of phonemic complexity examined. This variable was examined to observe the influence that frequency of occurrence of shared sounds had on the accuracy of the production of those same sounds. Fabiano-Smith and Goldstein (2010b) posed the following research questions.

1. Will bilingual children demonstrate evidence of interaction between their two languages as predicted by Paradis and Genesee (1996)?

2. Will bilingual children demonstrate evidence for a variation of the hypothesis of acceleration due to phonetic similarity (Flege, 1995)?

3. Does sound frequency predict high accuracy of shared sounds in the productions of bilinguals?

To answer these research questions, the authors recorded the speech of 24 children from Head Start programs in Philadelphia, Pennsylvania (monolingual English and bilingual Spanish–English), and preschool programs in Querétaro, Mexico (monolingual Spanish). Children were equally distributed into three groups: monolingual English (n = 8), bilingual Spanish English (n = 8), and monolingual Spanish (n = 8), with a mean age of 3;4 (years;months). Researchers used single-word and connected speech samples to derive measures of phonological ability. Single-word elicitation samples were taken using the phonology subtest of the BESA, a formal bilingual speech and language assessment for Spanish and English children (Peña et al., 2018). Connected speech samples were collected using a predetermined set of toys whose names targeted English and Spanish consonant sounds to elicit conversational speech. The connected speech samples were used to provide grammaticality scores to measure expressive language as well as supplement the establishment of the children's phonetic inventories. Authors reported the use of Spanish-speaking bilingual graduate students, but their involvement was limited to the phonetic transcription of Spanish samples (data processing) and not all were members of the Latinx community.

Evidence of transfer, acceleration, and deceleration was extrapolated from the data set. In summary, the authors interpreted the data in support of the dual-systems model, as all three hypotheses of between-language interaction were observed. Their first research question was supported by evidence of interaction between English and Spanish on measures of consonant accuracy (e.g., deceleration). Their second research question was supported by evidence from the monolingual Spanish-speaking group, in that they produced consonant sounds with higher accuracy as compared to the Spanish productions of the bilingual group. Finally, to answer the third question, statistical analyses were completed and showed that frequency of occurrence of sounds in English was not a predictor for accuracy of production of shared sounds of the bilingual group.

Fabiano-Smith and Barlow (2010)

Fabiano-Smith and Barlow (2010) investigated the acquisition of speech sounds in the phonetic inventories of Latinx bilingual Spanish–English–speaking preschoolers. To examine phonological acquisition, they studied phonetic inventories based on the hierarchical organization of speech sounds in a given language. Authors referred to two classes of sounds; marked sounds are relatively more complex sounds (e.g., early developing, high frequency of occurrence, and common across the languages of the world), whereas unmarked sounds are defined relatively less complex (e.g., late developing, appear infrequent, and are less common across languages). This hierarchy is further described in terms of levels with Level A being the least complex and Level E being the most complex. Although some sounds are generally considered marked and unmarked across languages, their place in a given typological hierarchy can differ cross-linguistically. For example, in English, the liquid /l/ is present at Level D (Dinnsen et al., 1990), whereas in Spanish, the liquid /l/ is present at Level E (Cataño et al., 2009). With this in mind, researchers asked the following questions.

1. Will phonetic inventories be identical for the bilingual children's two languages?

2. Will the phonetic inventories of bilingual children provide evidence for interaction between their two phonologies?

3. Will bilingual children have sparser phonetic inventories than their monolingual peers because they are learning two languages?

The same data set of 24 children, both bilingual and monolingual, as Fabiano-Smith and Goldstein (2010b) were examined in the work of Fabiano-Smith and Barlow (2010). Phonetic inventories were derived by analyzing single-word phonetic transcriptions, with connected speech samples used as supplements in the case of limited production opportunities available in the single-word sample. Sounds were considered part of the phonetic inventory if they appeared at least twice across the single-word and connected speech samples. Evidence of transfer was analyzed by noting any instance of speech sounds from one language appearing in the other inventory.

The authors observed bilingual children demonstrating phonetic inventory complexity comparable to their monolingual counterparts, in both languages. Furthermore, they found that the composition of inventories was similar, but not identical, across the two languages of bilingual children. In other words, sounds that appeared in both languages for some children may have appeared in only one language upon for others. These findings suggest evidence of separated English and Spanish phonological systems. Evidence of transfer was found when existence of English speech sounds was observed in the Spanish phonetic inventory of specific bilingual participants. In summary, their findings supported the dual-systems model in that linguistic systems were identified as separated in bilingual 3-year-olds but that bilinguals still produce speech sounds at age 3;0 in a manner similar to monolingual children. Further, their productions included a degree of transfer, indicating between-language interaction.

Fabiano-Smith and Bunta (2012)

Fabiano-Smith and Bunta (2012) investigated the role of voice onset time (VOT) in the speech production of Latinx bilingual Spanish–English–speaking preschoolers, within the theoretical framework of the dual-systems model (Paradis, 2001). The same data set examined in the works of Fabiano-Smith and Goldstein (2010b) and Fabiano-Smith and Barlow (2010) was used. Bilabial and velar stop productions shared between Spanish and English were observed to understand acquisition of VOT as well as the role of between-language interaction in acquisition of this phonetic cue. In the literature, evidence exists of differences in the acoustics of Spanish and English speech production. One difference is that singleton voiced stops in stressed syllables are prevoiced, while the same voice stops in English do not have prevoicing (Lisker & Abramson, 1964). Another point of difference is that voiceless stops, in the same position as the voiced stops previously mentioned, have a shorter VOT, whereas their English counterparts have a longer VOT. Within the dual-systems model framework, the authors summarize that the focus of the article was to investigate the acquisition of voicing contrast in bilingual Spanish–English–speaking children and the impact of cross-linguistic influence will have on development between monolinguals and bilinguals. Two research questions were proposed.

1. Do bilingual children produce VOT like their monolingual peers?

2. Are bilingual children differentiating between VOT contrasts or are they using the same categorical difference?

The authors communicated several findings. First, Spanish VOT values did not differ between the two groups, suggesting that early Spanish influence on English had more influence on bilabial stops than velar stops. Authors also found no significant differences between the VOT for Spanish and English stops between bilingual students, implying that bilingual children do not reliably differentiate the VOT of voiceless stops in Spanish relative to English. Results indicated an observed distinction between the voiceless velar stop in Spanish–English acquisition. In general, there were implications to suggest VOT as a site of cross-linguistic influence at the phonetic level. With this, Fabiano-Smith and Bunta (2012) concluded that their findings further support previous studies suggesting cross-linguistic influence in bilingual phonological acquisition (Fabiano-Smith & Barlow, 2010; Fabiano-Smith & Goldstein, 2010b; Paradis, 2001; Paradis & Genesee, 1996).

Review and Reinterpretation Criteria

Theoretical Framework Review

Original key interpretations of results within the dual-systems model framework, and the racial and ethnic profile of the research teams, were identified from each article and organized for each respective study. Results were then reinterpreted via the parameters of the translanguaging lens and the Insider researcher perspectives. To review, three parameters of the translanguaging lens include the following:

1. existence of a single unbound repertoire/idiolect;

2. language choices are the result of feature selection; and

3. repertoires may overlap between speakers.

The parameters of translanguaging were used to help guide authors in how findings and interpretations were framed. Authors from this study looked for any mention of separated phonological systems and discussions made by original authors on possible reasons for phonological errors. From this, the parameters of translanguaging assisted with describing the nature of the phonological system using the given findings as well as the phonological production behaviors of the participants.

Interpretations using the Insider researcher perspective were guided using the following questions: (a) Were any members of the research team members of the communities being studied (and/or was there no mention of identity)? (b) Did the interpretations include information on the sociolinguistic context of the community? (c) Did evidence of between-language interaction take into consideration pragmatic aspects of bilingual language use? (d) Were any productions counted as “errors” indicative of acceptable phonological productions by members of the community (even if outside the parameters considered “errors” by researchers, the research literature, or SLPs)?

Insider Reviewer

An Insider, for the purposes of this study, was defined as an individual who is Latinx and has spoken Spanish and English as heritage languages. Relating back to positionality, it is possible for insiders to be positioned outside a community (e.g., the first author would be an outsider if the study focused on undocumented individuals); positionality is dynamic and dependent on communities with which an individual works. The first author is a first-generation, Mexican Salvadoran individual. He grew up with both English and Spanish in majority of his social contexts. Additionally, he was raised in a Mexican and Salvadoran household with Insider perspectives of Latinx cultures. The first author spoke both named languages as a child but shifted to speaking majority English and only understanding Spanish through primary education. In secondary education, he began speaking Spanish with more frequency. As previously mentioned, authors of the articles disclosed that they utilized bilingual research assistants to help complete transcription and analysis. However, bilingual research assistants were not part of the study design, data collection, or data interpretation process. Upon review of these articles, the first author of this study interpreted all aspects of each study through the perspectives of an Insider. Given that the first author is positioned as an Insider for all three studies, the following interpretations were performed by the first author only.

Results

Summary of Reviews and Reinterpretations

Data were organized on tables by both the original study interpretations under a dual-systems model and the reinterpretations under a translanguaging lens. Table 1 responds to Fabiano-Smith and Goldstein (2010b), Table 2 responds to Fabiano-Smith and Barlow (2010), and Table 3 responds to Fabiano-Smith and Bunta (2012). To reiterate, one key difference between the dual-systems model and interpreting through a translanguaging lens is the concept between-language interaction; dual-systems model dictates a separation of systems, with some interaction, and the latter dictates a unitary system making up a linguistic repertoire containing features from which speakers can deploy. As a result, reinterpretations highlighted several differences of the findings that are discussed for each article.

Table 1.

Summary of Fabiano-Smith and Goldstein (2010b) interpretations and reinterpretation.

Original interpretations	Reinterpretations under translanguaging
When transfer occurred, as evidenced by stop consonant deaspiration, it was not random and it was systematic; therefore; there is a maintained separation of language and interaction at this level hardly occurs.	There was a selection of a feature that occurred because the child has available within their repertoire a deaspirated stop consonant. This suggests that the child can deploy a feature that still fits in appropriately (i.e., does not deploy a trilled “r”) suggesting the systematicity of feature deployment.
Bilingual children can use Spanish stops in English because those stops occur in English and can use them without changing meaning, producing a range of different phonetic variants for a single second language.	Bilingual children are deploying the correct linguistic feature as evidenced by the fact that they are not changing the meaning of the word. These are not variants of a named language, rather, this feature fits in the phonemic parameter enough to keep the same intended meaning. There are no “variants.”
Monolingual English children had higher accuracy on stops and fricatives than bilingual English children, supporting hypothesis of deceleration.	Under translanguaging, the idea that deceleration of one language relative to the other would not be considered valid. We cannot compare the linguistic abilities to a monolingual standard because that assumes a “monolingual-likeness,” which then separates out the named languages. Rather, we focus on the idiolect, the child's individual language and patterns of deployment. The bilingual child deploys features from their own linguistic repertoire that cannot be compared to that of a monolingual. Bilingual children may have formed a hybrid feature between stop sounds and fricative sounds or may deploy a different stop and fricative sound that still fits within the phonemic parameters of the intended target.
Twenty-five percent of bilingual children demonstrated transfer at a lower frequency between both languages; therefore, they maintain separation between their two phonological systems; how often transfer occurred provided little support for interaction between both languages.	Bilingual children deploy linguistic features at a lower frequency because of the nature of the feature deployed. They may deploy one feature over another not because of separation but because it may be more appropriate to use within the speaker condition. In other words, given the communication environment the child is in, the deployment of certain features at a lower frequency occurs because the other possible linguistic feature that could be selected may not be appropriate for the given interaction and not because there is a separation of systems.

Table 2.

Summary of Fabiano-Smith and Barlow (2010) interpretations and reinterpretation.

Original interpretations	Reinterpretations under translanguaging
The phonetic inventories of the bilingual children were equivalent to that of their monolingual peers in both languages.	This finding suggests that there is an overlap in the linguistic repertoires of the bilingual and monolingual groups.
The findings from this study that demonstrated the same bilingual children possess phonetic inventories that are like monolingual speakers of either language is evidence that a slower rate of development of one phonological skill does not indicate a slower rate of development for all phonological skills in bilingual children.	Bilingual speakers are utilizing different phonetic features in their linguistic repertoire that still fit in the phonemic parameters thus not changing the meaning of the word. Because they are selecting a different feature relative to their monolingual peers, it appears as if they are experiencing a slower rate of appearance.
That is, it may be that if a 3-year-old bilingual child demonstrates a sparse phonetic inventory in both languages, he or she may present with a speech disorder. Conversely, if a bilingual 3-year-old demonstrates phonological accuracy levels slightly lower than his or her monolingual peers, he or she may not require intervention services.	Under the scope of translanguaging, a speech-language service provider may look at phonetic inventory as a whole unified system rather than two separate phonologies. In other words, assessing what features are available that allow for functional communication. Through translanguaging lens, the focus would be on what phonetic features the child selects and if it is functional for their communication, not if the accuracy levels they achieve relative to monolingual levels.
The results from the analysis of inventory typology indicated that bilingual children organize their two speech sound systems in the same hierarchical fashion as monolingual speakers of English and Spanish.	A translanguaging approach would suggest that bilingual children organize their linguistic repertoire in a hierarchical fashion based on typology. Typology is recognized as a feature in a linguistic repertoire; therefore, the linguistic repertoire (i.e., phonetic inventory) can be organized according to this feature. Additionally, this organization is unique and, under a translanguaging approach, cannot be compared to that of monolingual repertoires.
Evidence of separation between the two languages of bilinguals was found through the examination of the levels of complexity in both English and Spanish. Most bilingual children were at Level D in Spanish but at Level E in English, demonstrating separation between their two phonological systems.	There is evidence that children deploy certain phonetic features in a given context over the other. The children of this study may have more experience utilizing certain phonetic features over others given their sociolinguistic contexts outside of the study. This would not be evidence of separation of phonologies but, through a translanguaging lens, evidence of the influence frequency of production of specific phonetic feature has on speech sound production in a given linguistic repertoire
Although these levels of complexity are similar, they are not identical; thus, bilingual children seem to recognize which phonological elements belong to Spanish and which belong to English and to organize them in a systematic but mostly separate manner.	These typologies are not identical, but the differences of productions based on typology are the result of the child's lived experience using their linguistic repertoire in their sociolinguistic environments. For example, these children may deploy sounds in more marked categories at higher frequency because those are the features shared by them and their interlocutor. In other words, the levels of complexity differ not due to separation of named languages, but rather the frequency at which they deploy the phonetic features.
Evidence of separation is clinically important in assessment and intervention because assessment and treatment of one phonological structure does not necessarily indicate assessment and treatment of that structure in the bilingual child's other language.	Separation of languages does not follow the perspectives of a translanguaging approach. Using the translanguaging approach, identifying a phonological system as unitary and unbound suggests that clinicians must look at the entire linguistic repertoire and its function in communication. Translanguaging bypasses the need to assess two separate systems and embraces the entire repertoire giving the child the opportunity to utilize their full linguistic system without boundaries.
During assessment, if a child possesses a sound his or her English phonetic inventory, it does not indicate that that same child possesses that sound in his or her other language.	This interpretation separates phonetic features into named English and Spanish. Through a translanguaging lens, a child may have, as well as deploy, a phonetic feature that is useful for their linguistic environment. Therefore, clinically, if the feature is used across contexts without breaking phonemic parameters, then this feature is available for functional communication.
In this study, evidence of transfer was found between English and Spanish in the phonetic inventories of bilingual children, indicating interaction between the two languages of bilinguals. Specifically, Spanish sounds were sometimes observed in the children's English inventories (e.g., Child B02: (b)) and English sounds were sometimes found in their Spanish inventories (e.g., Child B08: (a)).	It was found that children utilize phonetic features in a variety of contexts that allow the feature to be utilized without disrupting the communication signal. In other words, children were observed to have selected and deploy a specific phonetic feature from their linguistic repertoire that was functional for the given context.
These findings indicate that although bilingual children maintain separation for the majority of their phonological structures, there is a very low level of interaction between their two languages (see also Fabiano & Goldstein, 2005).	Interaction does not occur as it assumed delineation of language systems (i.e., two separate phonologies). Instead, these findings indicate that children have the ability to allocate phonetic features where they see fit. Better said, children were observed to use different phonetic features based on whom they speak to and the linguistic repertoires they know they share with the other interlocutor.

Table 3.

Summary of Fabiano-Smith and Bunta (2012) interpretations and reinterpretations.

Original interpretations	Reinterpretations under translanguaging
Spanish VOT did not differ between monolingual Spanish and bilingual Spanish, but Spanish influence on English is more evident on the bilabial plosive than the velar plosive.	Phonetic cues, such as voice onset time (VOT), are the same between individuals who share the same feature in their linguistic repertoire, but hybridization can occur between two features to create a novel feature that can be deployed.
The less marked (i.e., less complex) feature will be more easily transferred and persist longer in development.	Features that are less marked may be deployed more often because of their reduced complexity relative to the other features that can be selected to fit within the same phonemic parameters. Because of an increased deployment of this less marked feature, the child may utilize this phone selection for efficiency before acquiring a new phone that could be added to the repertoire and deployed.
Monolingual English VOT closer to adult than bilingual English, so bilingual acquisition of VOT might be slower than monolingual in this aspect.	In translanguaging, we focus on the idiolect; therefore, there is no function of a monolingual standard comparing VOT as a feature. Due to the nature of the preexisting VOT features that can be deployed, a bilingual child may deploy a hybrid, pre-existing, or better represented feature.
MS and BS demonstrated little variability for /p/ and /k/ due to preference for shorter lag time that tends to characterize early VOT development. This is evidence of cross-language influence due to equivalence classification.	The little variability is due to the overlap of features both groups share in their repertoires. Categories were not “blocked” and categorized into single groups, but rather, shorter lag times are a feature that exists in the bilingual Spanish group as well as the monolingual Spanish group; hence, it is likely to be deployed.
VOT is different in English but not in Spanish; therefore, phonetic level of speech production site of cross-language influence.	VOT is going to be different across stop sounds and the existence of two features can result in a blend or hybridization of a novel feature.

Review of Fabiano-Smith and Goldstein (2010b)

The key interpretation of findings in Fabiano-Smith and Goldstein (2010b) was between-language interaction as evidenced by higher accuracy on shared versus unshared sounds and systematic transfer of phonetic cues (e.g., aspiration). Under translanguaging, shared and unshared sounds as well as transfer (and even deceleration and acceleration), cannot occur because separated language systems are necessarily implied under the dual-systems model. Therefore, through a translanguaging lens, findings were reinterpreted as (a) shared sounds overlap with more idiolects in the child's linguistic environment, so children get more practice with those features, and (b) phonetic features (i.e., unaspirated stop consonants) occurred because they are features that exist within the child's linguistic repertoire and can be deployed at will. Specifically, the speaker could deploy a linguistic feature that fits within the given phonology, regardless of the named language category. While we agree with Fabiano-Smith and Goldstein (2010b) that these productions are systematic, the underlying organization that motivates this systematicity differs. In other words, the child deploys an unaspirated stop consonant because it is a feature within their repertoire rather than an element from one language being used in the other.

In addition, the authors originally interpreted findings that bilingual children used “Spanish” (i.e., unaspirated initial voiceless stop consonants) in “English” because English also uses unaspirated stops in an allophonic manner. More specifically, production of unaspirated versus aspirated stop consonants in English does not change the meaning of a word (i.e., not phonemically contrastive). As a result, bilinguals can utilize a range of different phonetic variants (i.e., stop aspiration) across both English and Spanish without disrupting the communication signal. This interpretation assumes separation of named languages and the interaction that occurs, whereas translanguaging asserts that bilingual children are deploying linguistic features that are different in production but do not change the meaning of the word. Moreover, VOT and aspiration would not be considered variants specific to a named language (e.g., unaspirated “Spanish /k/,” aspirated “English /t/”) since these features exist within the idiolect and can be deployed in linguistic contexts where their interchangeability is allowed. Thus far, we see that what was interpreted as transfer under the dual-systems model could be the deployment of an already-existing feature that the child deploys in a systematic and efficient way when reinterpreted through a translanguaging lens.

Fabiano-Smith and Barlow (2010) presented findings that, according to their interpretations, further support the dual-systems model. Several findings from this selected article made comparisons to monolingual groups; for example, they suggested that the bilingual phonetic inventory was equivalent to that of the monolingual inventory. However, applying a translanguaging lens challenges this idea because a bilingual comparison to a monolingual group inherently imposes a monolingual standard. Using a translanguaging lens reveals an overlap in linguistic repertoires between bilingual and multilingual groups, which explains the similarities in bilingual and monolingual inventories as seen in the original interpretations. Although the inventories of the bilingual children were equivalent to some degree, it was also argued that the similarities between monolingual and bilingual children were evidence for slower development of one phonological skill set. From a translanguaging standpoint, and that of raciolinguistic ideology, the original interpretation posits a deficit view of bilingual phonological acquisition. More specifically, bilingual children are developing slower than monolingual children by comparison. However, a translanguaging lens indicated that the reason for this appearance of slower acquisition was simply the result of the child selecting different phonetic features within their repertoire. Therefore, this is not evidence for slower acquisition but, rather, evidence of the child's experience with selecting and deploying different sounds given their sociolinguistic contexts.

In addition to comparisons to monolingual groups, other interpretations were connected to the concept of transfer (Fabiano-Smith & Goldstein, 2010b; Paradis, 2001). Much like the arguments of Otheguy et al. (2015), language systems are not truly separated if evidence of cross-linguistic influence (e.g., transfer) is presented. Fabiano-Smith and Barlow (2010) presented evidence of transfer as they argued for evidence of separation. Specifically, evidence of separation was present in the finding that bilingual children's phonetic repertoires existed at various levels of complexity (i.e., English at Level E, Spanish at Level D). When reinterpreted, the idea of hierarchical organization of phonetic features based on typological markedness does fit within the parameters of translanguaging, but the difference in complexity levels of phonetic inventories is better explained in terms of experience. Children may have more experience with marked sounds, or with unmarked marked sounds, given their sociolinguistic environment. To apply the example from Puerto Rican Spanish in the work of Fabiano-Smith et al. (2014), use of the Spanish trill /r/ may not be common in a particular linguistic environment. Rather, use of the flap /ɾ/ may be preferred.

Fabiano-Smith and Bunta (2012)

Recall that Fabiano-Smith and Bunta (2012) examined VOT values for bilingual children and compared those values to those of monolingual speakers of both languages. The major finding of this work was that Spanish VOT values did not differ significantly between the Spanish of the monolingual and bilingual children. Additionally, production data from bilabial plosives were extrapolated as evidence of Spanish influence on English. In the reinterpretation, we argue that evidence of differing productions of bilabial plosives may imply interactions between features within an idiolect to form a novel feature. Similarly, researchers found that monolingual English VOT was closer to monolingual adult productions compared to bilingual English VOT. This was originally interpreted as a slower rate of acquisition in terms of VOT. In a different light, there cannot be a comparison to a standard (i.e., monolingual adult productions) when applying a translanguaging lens. This is rooted in the idea that translanguaging emphasizes an individual's unique linguistic repertoire, or idiolect. The difference in VOT, much like that of the previous interpretation, is due to a hybrid feature being formed or there exists a better represented feature in the repertoire.

Fabiano-Smith and Bunta (2012) found that monolingual and bilingual Spanish productions of voiceless bilabial and velar plosives showed little variability due to the preference for shorter lag time. In some contexts, English voiced and Spanish voiceless plosives both have a shorter lag time (Flege, 1995). Interpreted differently, translanguaging argues that the lack of variability in VOT is due to the overlap in repertoire of both groups. Additionally, shorter lag times can be thought of as a feature within the child's idiolect and are therefore more likely to be deployed.

Moving toward findings on markedness, researchers interpreted results as less marked (i.e., less complex) features, like stop sounds, are more likely to be transferred between languages. The original interpretation implied separation of systems, as it frames this phenomenon as transfer of features between the child's phonological systems. Applying a translanguaging lens, this process could be interpreted as less marked features are selected for production more frequently due to their reduced complexity (compared to other features) and can fit within the same phonemic parameters (i.e., to not change meaning). Because these less complex features are used more frequently, phone selection may occur to make acquiring a new phone more efficient, that it could be added to the repertoire and deployed with greater ease. Relative to adults, children may use this less complex feature more frequently because it is earlier acquired during development and therefore present in the repertoire before developing more marked (i.e., more complex) sounds in an adult repertoire.

From the Insider Perspective

Across all original articles, sociolinguistic contexts were not considered when data were interpreted. More specifically, effects that could have occurred as a result of the social context of the communication exchanges in the child's environment were not included in data extrapolations. For example, in the reinterpretations of Fabiano-Smith and Goldstein (2010b), the Insider (i.e., the first author of this study) noted that certain sounds may have been produced given the social context of school. One lived example from the first author is that they may pronounce the area in Tucson, Arizona, known as “Tanque Verde” as [ʈɑŋkə vɝdi] when speaking with individuals that share one set of features (i.e., bilingual Spanish–English speakers), as it is a closer production to the word in the language of origin. The production [teɪŋkə vɝdi] is what the author might select (to increase clarity) when speaking with individuals with a less diverse set of features (i.e., monolingual English speakers). Furthermore, the feature selection based on communication partner may have been misinterpreted as an error, motivating “deceleration” in acquisition. Insider knowledge illuminates that this production could have been attributed to deployment of one set of features as the result of the child recognizing the overlap in features with their interlocutor. It is possible that this interpretation was omitted due to the lack of awareness of language being used as a social tool and the pragmatics that guide linguistic choices within a given community.

Another Insider observation was the harm caused to the readership by the use of monolingual groups for comparison. For example, support for the hypothesis of deceleration was evidenced by monolingual English speakers producing stops and fricatives at higher accuracy relative to bilingual English speakers (Fabiano-Smith & Goldstein, 2010b). As an Insider, this finding, worded in this way, can communicate to members of the Latinx community that monolinguals are “better” at speech, that bilinguals are slower to acquire sounds than they would if they spoke only English. Another example of this issue can be found in the work of Fabiano-Smith and Bunta (2012): Monolingual English VOT better approximated adult production values as compared to bilingual English VOT. This comparison also furthers the narrative that bilinguals are less accurate, or nonstandard. Once again, there is the assumption that acquisition of these phonetic features is “better” if the child spoke only English or that monolingual production is the “gold standard.” While there is still emphasis on still being within expected limits of acquisition, this then raises questions surrounding the need for comparison to a monolingual standard if bilingualism does not cause delay. As emphasized by work in raciolinguistics, implications of these comparisons continue to attribute prestige to monolingualism.

Discussion

Research Questions Revisited

After completing the reinterpretations of the three research studies, the authors of this study were able to identify possible implications for each of the research questions posed:

Question 1: What linguistic phenomena can be uncovered when shifting data interpretation from a dual-systems model framework to a translanguaging lens?

To reiterate, one key difference between the dual-systems model and a translanguaging lens is the construct of linguistic systems. The dual-systems model states that two separated but interacting systems exist in the bilingual child (Fabiano-Smith & Bunta, 2012; Fabiano-Smith & Goldstein, 2010b; Paradis, 2001; Paradis & Genesee, 1996). Translanguaging argues that the there exists a single unified system as an idiolect where individuals can deploy linguistic features out of their unique repertoire (Otheguy et al., 2018).

One major difference in theoretical interpretation, across all three reviewed studies, was the concept of between-language interaction. In the work of Fabiano-Smith and Bunta (2012), there was evidence of interaction between Spanish and English as it appeared that Spanish's influence on English manifested as observable differences in productions of bilabial plosives. However, after reinterpretation, it was viewed that “influence” (i.e., movement of a phonological element from one inventory to another) could not take place if all features were already present in an unbound repertoire. Furthermore, VOT was no longer viewed as features of the named languages “Spanish” and “English” but rather variations of one phonetic feature in an idiolect. As a result, reinterpretation of findings did not explain these phenomena as interaction between languages but rather sets of features that speakers can select from and deploy when they deem appropriate (e.g., deployment does not change meaning; child shares features with some interlocutors but not others). Another finding that arose from this study was the idea of hybridization. Once more, because interaction between two language systems cannot exist within a single idiolect, differences in VOT were better understood as interactions between features within a given repertoire, or hybridization of features. This was also the interpretation in the work of Fabiano-Smith and Goldstein (2010b), which suggested that the differences in acquisition were the result of a slower acquisition process in the bilingual group, where translanguaging offers that those differences in production are a result of differences across individual idiolects. Monolinguals also possess their own individual idiolects; thus, group comparisons (i.e., bilingual vs. monolingual) are not feasible. Fabiano-Smith and Barlow (2010) added insight to the discussion on phonological transfer (i.e., cross-linguistic interactions). Transfer was evidenced by children using sounds from one named language in the other. However, like in the work of Fabiano-Smith and Goldstein (2010b), translanguaging would argue that transfer cannot occur if two separate languages do not exist. Otheguy et al. (2015) would also question if systems were truly separated if they were interacting.

Question 2: Does the community-based wisdom of an Insider researcher uncover linguistic phenomena in previously reported findings by all-white research teams?

As stated above, an Insider researcher is an individual who shares a cultural connection with the community being studied (i.e., positioned within the community). Therefore, it should be stated that simply being bilingual in the named languages studied is not enough to achieve accurate and culturally representative interpretations. As an Insider, two key notions were challenged across these three articles: sociolinguistic context and reference to a monolingual standard.

In terms of sociolinguistic context, the reinterpreted articles generally did not acknowledge the role of speaker context. Speakers who are bilingual and carry strong cultural ties to the communities with which their language is also associated often utilize their repertoire in a pragmatic manner. Specifically, they often communicate to build social connection rather than in a formal, school-based testing context. Insider speakers can identify where an overlap in repertoires may occur in conversation and may use additional features (Otheguy et al., 2015). Many differences that were observed across the three studies between monolingual and bilingual groups may have been influenced by the contexts of the data collection process. These contexts may have included the child's perception of the researcher, the child's readiness to participate, and the child's perception of whether formal or informal language was expected in the data collection setting. For example, the Insider author (i.e., first author) navigated a predominantly white elementary school as a child. This meant that the use of specific linguistic features (that followed a monolingual template) was expected for interactions with teachers, teachers' aides, and other administration. When in the presence of other Latinx students, the first author was able to utilize a distinct set of features. What is important to point out is that the Insider knew when they were able to use their full repertoire. They could recognize that the Latinx students could understand and respond pragmatically during communication exchanges. This sociolinguistic phenomenon could have been part of the data collection design to control for children using different features with different interlocutors.

On the topic of Outsider perspectives, there is a lack of lived experience and, as a result, understanding about the role languages play in minoritized communities when research teams do not include members of the community being studied. Very rarely is this lack of lived experience ever questioned in the research community. Insiders provide the cultural knowledge and further detail on the sociolinguistic contexts in which communication occurs. For example, Insiders may point out to a research team that children may take on prosodic and phonological features that increase intelligibility for their interlocutor. When speaking to individuals inside their community, however, they may switch features. For example, children may employ a faster rate of speech, deaspiration on initial voiceless stops in English, or the use of community-specific words or phrases. Furthermore, the research team can benefit from the deeper cultural connection to minoritized communities. Simply being linguistically bilingual but carrying no cultural ties to the communities under study may limit data interpretation, resulting in incomplete findings.

Another common research method that promotes deficit model thinking (Harry & Klingner, 2007) is the use of monolingual comparison groups, specifically, predominantly white, monolingual English samples. The Insider reinterpretation of this method was that bilingual–monolingual comparisons stigmatize bilingual speakers (De Houwer, 2022). Verbiage such as, “when compared to monolinguals,” can be interpreted by members of the Latinx community that monolinguals are the standard, and bilinguals must meet that standard. Connecting back to raciolinguistics, we emphasize that raciolinguistic ideologies emphasize that the language practices of racialized communities are inherently deficient (Flores, 2020). Additionally, being that their samples tend to use the white variant of English, this compounds the raciolinguistic ideologies in that bilinguals are presented to have deficits in their language abilities when compared to the white, English standard created by monolingual English comparison groups. The authors of many all-white research teams strongly advocate that bilingualism does not cause harm during language acquisition; however, utilizing language like “slower/lower than monolingual groups, but within normal limits” sends the message, whether intentionally or unintentionally, that bilinguals are not performing at age level; they are deficit in their communication. The language of “lower” or “sparser than” helps to perpetuate the mythology promoted by pediatricians, teachers, and SLPs: that parents should abandon their heritage language and speak to their children “only in English” if their child exhibits a speech-language disability.

In general, the findings of our reinterpretations were heavily related to the sociolinguistic aspect of language and languaging. Across all three studies, there was a lack of acknowledgement and understanding of how the social context of communication can impact how bilingual individuals utilize their repertoire. Moreover, because of the dynamic nature and robustness of their idiolect, it does not make sense to use monolingual groups as “controls” for bilingual studies. Without a doubt, bilingual speakers are not two monolinguals in one (Meisel, 2006), so continuing to compare bilingual languaging to monolingual languaging, especially to the white variety as mentioned above, is problematic for both research and the communities being studied.

Conclusions

The purpose of this study was to organize interpretations made using the dual-systems model and reinterpret them using translanguaging as a lens through which we may view data interpretation. In addition, we attempted to understand the impact that Insider perspectives have on the interpretation of data in bilingual phonological research. We shared the perspective that translanguaging could lend to the field of communication sciences and disorders. It reinforces the idea that speakers have their own individual repertoire from which they select features. In combination with the novel theoretical framework, the Insider perspectives brought to light the social, humanistic nature of feature selection during communication.

Translanguaging challenges the use of language standards, allowing for individuals to take language as their own lived experience, incomparable to any other experience. One last point that echoes throughout the discourse of linguistic diversity and translanguaging is that we do not utilize translanguaging as the means to view bilingualism/multilingualism as “good” but to view bilingualism/multilingualism as the standard.

Funding Statement

The authors would like to thank the children, families, and communities that participated in the original studies reviewed in this review article. Graduate research assistant funding for the first author was provided by the National Institute for Deafness and other Communication Disorders under Grant 5 R01 DC016624-04 (awarded to Leah Fabiano-Smith, principal investigator).