A Qualitative Analysis of Translanguaging by Colombian Migrants in North Carolina

: This study examines the usefulness of the theoretical construct of translanguaging in analyzing the linguistic production of twenty-four Colombians (originally from Bogot á ) residing in the Piedmont Triad region of North Carolina. Translanguaging maintains that bilinguals and multilinguals have a single linguistic repertoire consisting of features traditionally associated with different named languages (English, Spanish, Arabic, Chinese, etc.), and that they freely select from among these features according to their communicative needs in speciﬁc contexts. In terms of named languages, participants utilized varying amounts of English during sociolinguistic interviews conducted primarily in Spanish by the investigator. The study presents a qualitative analysis of participants’ linguistic production viewed through the lens of translanguaging, which offers a better account than codeswitching of several patterns of language use observed in the data. These include phonetically ambiguous words, ﬂuid combinations of morphemes from each named language (Span-ish and English), and innovative uses of linguistic forms. The study concludes with a consideration of the relevance of translanguaging in addressing the issue of the legitimacy of the (often stigmatized) language varieties of Hispanics in the U.S. context.


Introduction
The phenomenon of multilingualism is a noteworthy feature of our contemporary globalized world, as evidenced by its growing prominence within the field of linguistics (Auer and Wei 2007;Bhatia and Ritchie 2013;Martin-Jones et al. 2012). More specifically, the intersection between multilingualism and migration has proved to be a fruitful area of research in recent years due to the exponential increase in mobility both of individuals and of the linguistic resources they employ (Blackledge and Creese 2017;Blommaert 2010;Canagarajah 2017;Duarte 2020;Gal 2006;Keating and Solovova 2011;Martin 2007;May 2014).
In light of this growth in the mobility of human populations, linguists have sought to describe how migrants employ their linguistic repertoire (including elements classified as belonging to different languages) to meet their communicative needs in the novel social and linguistic contexts in which they find themselves. Translanguaging (García 2009;García and Wei 2014;Vogel and García 2017), which is a theoretical approach to multilingualism originally applied to language pedagogy as a way to leverage the full linguistic resources of multilinguals for teaching and learning, also lends itself well to the study of language in migrant contexts. Translanguaging "posits that rather than possessing two or more autonomous language systems, as has been traditionally thought, bilinguals, multilinguals, and indeed, all users of language, select and deploy particular features from a unitary linguistic repertoire to make meaning and to negotiate particular communicative contexts" (Vogel and García 2017, Summary section, para. 1). The central focus of translanguaging is the idiolect, defined by Otheguy et al. (2015, p. 289, emphasis in the original) as " . . . a person's own unique, personal language, the person's mental grammar that emerges in interaction with other speakers . . . ", rather than socially constructed named languages such as Spanish or English. Translanguaging thus conceptualizes the linguistic production of bilinguals and multilinguals in a different way than codeswitching, which presupposes the existence of two or more separate linguistic codes with speakers alternating between (or among) them during the same conversation (Bullock and Toribio 2009;MacSwan 2014;Milroy and Muysken 1995;Poplack 1980Poplack , 1988Toribio 2011;Torres Cacoullos and Travis 2018 1 ). Whereas codeswitching requires identification of specific linguistic features (phonemes, words, phrases, etc.) as instances of 'language X' or 'language Y', translanguaging requires no such categorization. Instead, the theory focuses on how speakers utilize features from their idiolect to express their intended messages, influenced "...by the social information that each individual speaker has regarding the particular communicative context in which the social interaction takes place" (Vogel and García 2017, Debates in the Field section, para. 2). It should be apparent from the above description that the role of the speaker, the listener, and the context of interaction are all important to consider when adopting a translanguaging framework.
The aim of the current study is to examine the usefulness of translanguaging for analyzing the linguistic production of twenty-four Colombian migrants residing in the Piedmont Triad region of North Carolina, an area with a relatively recently established (majority-Mexican) Hispanic community (U.S. Census Bureau 2010 2 ). This study is the first to examine translanguaging among Colombians in the United States. While the amount of research on Colombian Spanish in the U.S. has grown in recent decades (Hurtado 2001(Hurtado , 2005a(Hurtado , 2005bHurtado Cubillos 2012;Lamanna 2012aLamanna , 2012bOrozco 2004Orozco , 2007aOrozco , 2007bOrozco , 2012Orozco , 2015Orozco , 2018aOrozco , 2018bRamírez 2003Ramírez , 2007aRamírez , 2007bVélez-Rendón 2014), it remains relatively understudied in comparison with other U.S. Spanish varieties, especially Mexican, Puerto Rican, and Cuban Spanish (see Lipski 2008 for an extensive bibliography). Colombians in the United States have an interesting status as a minority within a larger minority language community, since they do not represent the majority in any of the main Spanish-speaking communities in the U.S. (Lamanna 2012a). They are also speakers of a relatively prestigious variety of Spanish vis-à-vis other varieties, and thus enjoy a high degree of linguistic security (Zentella 2002). The interplay of these two factors influences how members of this community index their identity as Colombians and/or as members of a larger multiethnic Hispanic community through their use of specific linguistic forms, including those traditionally associated with English. The findings of the current study indicate that translanguaging offers a more accurate and nuanced account than codeswitching of several patterns of language use observed in the data from this community, as explained in further detail below.

Materials and Methods
Twenty-four Colombians (who were all residing in the North Carolina Piedmont Triad at the time of data collection) participated in this study. Subjects were either originally from the Colombian capital city of Bogotá or had lived there most of their lives prior to migrating to the United States. They included thirteen males and eleven females and ranged in age from 18 to 65 (mean age = 44.4). All participants had either attained or begun their secondary or post-secondary education, with the majority having attained or begun the latter. Participants were limited to those who had resided in the Piedmont Triad area for at least two years, with length of time in the community ranging from 2 to 9.5 years (mean length of residence = 5.92 years).
Members of this migrant community maintain a strong identity as Colombians through their participation in transnational social spaces (Faist and Özveren 2004). Faist (2004, p. 3) defines transnational spaces as "relatively stable, lasting, and dense sets of ties reaching beyond and across the borders of sovereign states". The participants in this study maintain contacts with the home country via technology and by travelling between the United States and Colombia (some frequently, others less so). Their Colombian identity is also reinforced through participation in informal get-togethers (parties, etc.) with other members of the Colombian migrant community. These gatherings also sometimes include individuals from other Spanish speaking countries who currently reside in North Carolina (mainly Mexicans, but also migrants from places such as Costa Rica, Puerto Rico, etc.).
The investigator recruited subjects primarily through a social network approach (Milroy 1987;Milroy and Llamas 2013). Recruitment (which occurred in person, over the telephone, or via email) began with the investigator's personal contacts in the area and subsequently included individuals in the contacts' social networks (family and friends). Subjects who were minors were recruited through their parents rather than contacted directly. Participants were also recruited through flyers posted in areas where potential subjects were likely to be found (Hispanic stores, churches, universities, etc.).
Participants were interviewed between June 2008 and January 2009 by the investigator, a native speaker of English who is also fluent in Spanish. A range of topics was addressed in the questions (see Appendices A and B), including elements of the traditional Labovian sociolinguistic interview such as relating childhood memories and describing a near-death experience (Labov 1984;Tagliamonte 2006). Other questions were included to obtain information relevant to a study of language and dialect contact, such as the informants' language use (Spanish and/or English) in different contexts as well as their contact with and attitudes towards other Spanish-speaking groups. Participants frequently provided information specifically about their experiences with Mexicans and/or Mexican Spanish in response to these questions. This information, however, was not gathered systematically during the interviews, and as a result the amount of time spent answering the questions varied from participant to participant.
The investigator digitally recorded participants' responses on a laptop computer using the sound-recording software Audacity and a microphone. Data collection took place in participants' homes whenever possible, to protect their privacy. When this was not feasible, data were collected at a mutually convenient location preferred by the participant, such as his or her place of employment. Although one disadvantage of the above approach is the presence in some of the recordings of background noise typical of residential settings (water running from a faucet, voices of other family members, etc.), such ambient noise rarely interfered with the ability of the investigator and/or the research assistants who transcribed the data to understand the language produced by the informants.
The interviews were transcribed by research assistants who are native speakers of Colombian Spanish. The method of transcription employed was that described in Bentivoglio and Sedano (1993), with a few minor modifications. Transcriptions were subsequently checked by the researcher for accuracy.
Although the investigator conducted the interviews in Spanish, and participants were told that they were participating in a study on Spanish in North Carolina, they were not explicitly instructed to avoid using English in their responses, and they all spontaneously produced at least some linguistic features traditionally categorized as English.
A qualitative methodology (Lew et al. 2018) was chosen for the current study to allow for an examination of how participants' translanguaging practices serve specific communicative ends in the context of an interview. A quantitative analysis of the data was not carried out due to the relatively small number of cases involved. The results of the analysis are reported in the following section.

Results and Discussion
Several categories emerged from an examinaton of the instances of translanguaging found in the interview data 3 . While not all speakers produced tokens representing each category, there was a considerable amount of overlap in categories among speakers, with no differences noted due to age, gender, or length of residence in the US. The relative uniformity observed may be due to their regular contact with Colombia and with other members of the Colombian migrant community, and/or result from the fact that they were all addressing the same interlocutor (the interviewer).
The categories appear below with an illustrative example for each, selected because it clearly represents the category in question. In this example, the words socially categorized as English (North Davie, South Davie) merely serve the function of identifying the schools. In a translanguaging framework, they do not need to be analyzed as switches from one language to another within the longer string of discourse, but rather simply represent cases of the speaker selecting the appropriate items from his linguistic repertoire to refer to these places.

Referents Salient in the U.S. Context but Not in Colombia
This category includes objects, concepts, activities, and events that are salient within the cultural context of the United States, but either unknown or uncommon in Colombia. In Example (2) below, the speaker uses the word tenure to refer to permanent employment as a professor at a U.S. college or university.
(2 Although the lexicon of standard Spanish includes a word that refers to the same concept (titularidad), the more salient word for this speaker is tenure, due to her experiences as a graduate student and subsequently as a professor in the United States. She also assumes (correctly) that the interviewer will understand her due to his own experiences within academia. For these reasons, tenure is the more appropriate choice to express her intended meaning and communicate efficiently and effectively within the specific context of this interview.

Representations of Other People's Speech
Participants in this study sometimes crossed named language categories in order to imitate the speech of another individual, as in example (3)  In this example, the speaker employs elements of his linguistic repertoire associated with English to demonstrate how a specific type of person (a typical student at the University of North Carolina at Chapel Hill) speaks. An attempt to render the same (or similar) meaning with linguistic features associated with Spanish would not have achieved the desired effect of representing the speech of a typical UNC student as accurately as possible. The speaker has therefore chosen the most appropriate items from his linguistic repertoire to achieve his communicative goal.
3.1.4. Discourse Markers Matthews (2007, p. 108) defines a discourse marker as "(a)ny of a variety of units whose function is within a larger discourse rather than an individual sentence or clause". Subjects in the current study engaged in translanguaging with a range of discourse markers such as so, I mean, and anyway (as in Example (4)  Similar examples have been analyzed in previous studies of Spanish-English bilingual speech (e.g., Lipski 2005;Torres 2002), and one of the issues addressed is whether they should be considered lexical borrowings or single-word codeswitches (a determination which is often difficult to make). This debate is irrelevant within a translanguaging framework, since the focus is on how the item is used in a given context, in this case as self-digression management (Ferrara 1997) while responding to a question during the interview.

Adjectives
Many of the examples of translanguaging found in the data involve adjectives. In these cases, the speaker chooses to cross a named-language boundary to express the precise meaning he or she wishes to convey. Despite the existence of dictionary equivalents to mellow in standard Spanish, none communicates the exact range of nuances of meaning conveyed by mellow. In this case, the choice of mellow may also have been intended to make an explicit connection with a restaurant previously mentioned in the discourse, the Mellow Mushroom.

Acronyms
It was common for speakers to render acronyms using the pronunciation of the letters in the named language associated with the acronym, regardless of the surrounding linguistic context. Here, the speaker refers to Guilford Technical Community College by its commonly used acronym, GTCC, which functions as a proper noun referring to the school.

Miscellaneous
Many instances of translanguaging in the data did not fit into any of the above categories. Example (7) below is typical, with one word in a list produced with features typically classified as English, and the rest with Spanish features.

Explanatory Value of Translanguaging
The data include numerous instances of language use where a translanguaging approach offers more explanatory value than an analysis framed in terms of codeswitching. These include words that resist clear phonetic classification as either "English" or "Spanish", combinations of morphemes from these two named languages that show no apparent regard for any distinction between them, and the use of specific forms in novel ways that differ from how they are typically employed in either language.

Words of Ambiguous Phonetic Classification
In the case of some place names, a word was pronounced partly with features associated with English, and partly with features associated with Spanish. For instance, in Example (8)  Translanguaging accounts for examples like this one better than codeswitching because the investigator does not have to decide if the language sample counts as "English" or "Spanish". The emphasis with translanguaging is on the word itself and how it is used, rather than determining the named language category to which it belongs. The existence of such examples requires no special explanation, since the speaker is simply combining elements from his unitary linguistic repertoire (in this case, phonetic features) that would be categorized by society as belonging to different languages. In fact, we would expect and even predict the occurrence of such examples, if speakers are indeed not keeping the languages neatly separate in their brains.

Fluid Combination of Morphemes from Each Named Language
Example (9) below involves a seamless mixture of words traditionally categorized as either English or Spanish in a description contrasting rush hour traffic in Miami with traffic in North Carolina. 4 An anonymous reviewer pointed out that there is a preference for shorts (pronounced [t or(es)] over pantaloneta or pantalones cortos among middle-aged speakers (40-60 years old) from Bogotá. This preference notwithstanding, Participant 7F1C53 in the current study clearly pronounced this lexical item using phonemes associated with English [ c r ts]. This example perfectly fits Vogel and García's description of translanguaging as involving speakers who "fluidly use their linguistic resources-without regard to named language categories-to make meaning and communicate" (Vogel and García 2017, Origins of the Term section, para. 4). The fact that the speaker uses morphemes traditionally associated with different languages within the same syntactic unit (the noun phrase las highways and the phrasal adjective bumper a bumper) is evidence of his disregard for named language categories as his language practices "go beyond" them, thus exemplifying the meaning of the transprefix in translanguaging (García and Wei 2014). Attempting to account for this example in terms of codeswitching requires asserting that the speaker switches from English to Spanish at three consecutive morpheme boundaries when uttering es bumper a bumper. Viewed through a translanguaging lens, the speaker is simply selecting appropriate items from his unitary linguistic repertoire to negotiate meaning during communication.

Novel Uses of Linguistic Forms
A number of cases observed in the data involve uses of specific linguistic forms in ways that notably differ from how they are employed in either standard English or Spanish, thus reflecting each speaker's unique idiolect 5 . Example (10)  Another example of speaker creativity involves a novel usage of the adjective down to refer to a city without much activity. In Example (11)  This speaker appears to be taking as a starting point the use of the word down to refer to a person who is sad or depressed, and applying it to a city that he and his family found especially unexciting. From a translanguaging perspective, speaker agency is driving this innovative use of an element from his linguistic repertoire, and the speaker is not at all constrained by how the word is generally used by others.
The cases of translanguaging described above demonstrate how Colombian migrants employ linguistic features as resources for effective communication in an interview context, with little to no regard for named language categories. They thus provide empirical evidence for the utility of concentrating on resources rather than languages in studies of bilingualism and multilingualism. Blommaert (2010, p. 21) argued that the focus in sociolinguistic research in an era of globalization needed to shift from languages to resources, and asserted that " . . . mobility is a central theoretical concern in this sociolinguistics of resources". As speakers move from one location to another they bring their preexisting set of linguistic features with them, acquire new ones, and skillfully deploy both as they adjust to and continue participating in the novel environments in which they find themselves. A translanguaging approach readily accounts for this fluid use of resources due to the central role played within the framework by "speakers' own dynamic linguistic and semiotic practices" (Vogel and García 2017, Theoretical Foundations and Assumptions section, para. 1).

Conclusions
The purpose of the current study was to apply the theoretical framework of translanguaging to an analysis of the linguistic production of a group of Colombian migrants in North Carolina, in order to assess its explanatory value vis-à-vis the more traditional perspective of codeswitching. Qualitative analysis of the data suggests that translanguaging does in fact provide a more accurate and nuanced account of the patterns of language use observed in this community than codeswitching, due to the difficulty of classifying certain elements phonetically as clearly "English" or "Spanish", the free and fluid combination of morphemes from these two named languages, and the fact that speakers sometimes employed specific linguistic forms in ways that diverge from those observed in either standard English or Spanish.
Future research should apply a translanguaging framework to study the linguistic production of additional understudied Spanish-speaking communities in the United States. Data should be collected from speakers representing a range of national origins (e.g., Central Americans, other South Americans) and geographical regions of the country, to determine if similar findings to those of the current study are obtained. The effects of different social variables such as age and gender on translanguaging should also be examined, using a larger data set with a sufficient number of cases to allow for statistical analysis. Another potentially fruitful area to explore is the relationship between speakers' linguistic repertoires and specific communicative events during which they are employed, going beyond the interview context included in the current study. Finally, translanguaging data should be obtained from interactions carried out via various technological platforms in addition to in-person conversations.
On a final note, translanguaging can be profitably employed to argue for the legitimacy of the language varieties of U.S. Hispanics, which unfortunately are often stigmatized by members of other groups (Otheguy 2009;Zentella 2007). Instead of depicting speakers of these varieties as deficient in some way, translanguaging sheds light on how they skillfully select features (traditionally associated with Spanish and/or English) from their linguistic arsenal according to their communicative needs and goals at a particular moment, and often in innovative ways. Translanguaging resists monolingual language ideologies that would insist that bilinguals limit their linguistic production to one named language or the other, allowing them to flourish instead (as clearly seen in the examples from the current study) by leveraging their full linguistic repertoire during communicative interactions. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Indiana University Bloomington (IRB Study #08-13186, approved 5/31/08).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to the need to protect participant confidentiality.