For reasons that touch on both linguistic theory and educational practice, recent scholarship has generated lively debate on the nature of the linguistic system that enables speech and writing in people who are described as bilingual or multilingual, a debate articulated in terms of the different positions taken with respect to the notion of translanguaging. Defined by Otheguy et al. (2015: 281) as “the deployment of a speaker’s full linguistic repertoire without regard for watchful adherence to the socially and politically defined boundaries of named (and usually national and state) languages”, translanguaging has disrupted conventional concepts of bilingualism and multilingualism as simply the mastery of two or more languages from birth or as a result of an additive process. (In our definition of translanguaging we use named language to refer to such social categories as ‘Arabic’, ‘Bulgarian’, ‘English’, ‘French’, ‘Igbo’, ‘Spanish’, ‘Swahili’, etc.; we owe the term to Makoni and Pennycook 2007). Together with many teachers and administrators, scholars in educational theory have actively engaged with the concept of translanguaging, which has led to pedagogies alternative to the traditional monolingual ones in second language, bilingual, and foreign language classrooms (García and Kleyn 2016; García et al. 2017; García and Wei 2014; García and Wei 2014; Gort and Pontier 2013; Gort and Sembiante 2015; Mazak and Carroll 2017; Paulsrud et al. 2017). But scholars in grammatical theory have so far paid scant attention to translanguaging, consistent with the tendency to disregard questions of how theories of language impact the lives of actual speakers, especially minoritized ones. We thus regard it as a welcome development that the translanguaging debate has attracted theoretical linguists like Jeff MacSwan and two of the present authors (Otheguy and Reid), and that one of the questions most controversial in the debate has been the theoretical one regarding the conception of the system underlying linguistic behavior in general and translanguaging behavior in particular.
To anticipate the central point of this paper, two conceptions of the linguistic system of bilinguals are in contention. In adopting the notion of translanguaging, Otheguy et al. (2015) espouse what we call a unitary view, arguing that bilingualism and multilingualism, despite their importance as sociocultural concepts, have no correspondence in a dual or multiple linguistic system. The myriad linguistic features mastered by bilinguals (phonemes, words, constructions, rules, etc.) occupy a single, undifferentiated cognitive terrain that is not fenced off into anything like the two areas suggested by the two socially named languages. In contrast, MacSwan (2017), in a strong critique of our work, has risen to the defense of the familiar position that, despite some overlap, bilinguals possess two separate linguistic systems that largely correspond to, and whose boundaries coincide with, those of the two named languages. The debate, then, is about the psycholinguistic reality of the two socially named languages of the bilingual. We maintain the position that, given the available evidence, there is no reason to believe that there is a cognitive duality correspondending to the social duality; MacSwan, with some qualifications, maintains that there is. The present paper offers a detailed rebuttal of MacSwan’s critique and a defense of the notion of translanguaging and the unitary view. And it spells out again the reasons for our conviction that the terms monolingual, bilingual, and multilingual make reference to sociocultural constructs and boundaries that tell us but little about the constructs and boundaries of the underlying linguistic system.
MacSwan’s critique tends to confuse matters by himself adopting the term translanguaging. But as we show in this paper, his ideas about translanguaging differ greatly from those of of other scholars. MacSwan insists on the need to conceive of the languages of the bilingual as internally differentiated and for the differentiation to be language-specific. And he persists on establishing this internal differentiation, despite claims to the contrary, on the basis of the named languages. We use translanguaging only in its original sense, namely that of promoting forms of theorizing about the linguistic system underlying the behavior of bilinguals that go beyond the limitations imposed by named languages. The trans- in translanguaging is taken in the sense of ‘to go beyond’ (Li Wei 2011), as in transcend). In this original sense, translanguaging offers a different ontology for the languages of the bilingual. It registers (and celebrates!) the separate existence of the two languages as important social realities, ones often most relevant to serious and heartfelt vindication struggles. But it proposes that the social duality is not matched by a dual psycholinguistic one.
MacSwan (2017) voices a serious and highly commendable concern with the effects that theories of bilingual competence can have on the well-being of members of linguistic minorities. We have ourselves devoted long years to studying, supporting, and developing educational programs for bilingual and otherwise linguistically minoritized communities. And Otheguy and García are bilingual members of the U.S. Latino community (with bilingual parents, children, and grandchildren). We are thus in strong sympathy with MacSwan’s enlightened position on the social front; our translanguaging approach does not in any way negate the value and importance of bilingualism and multilingualism for speakers and communities worldwide, including our own. Far from it, the translanguaging model supports the desires of speakers to act socially as bilinguals and multilinguals, to have two or more cultural and socio-political identities. But we differentiate, in ways that MacSwan and others fail to do, between the external sociocultural construct of named languages around which identities might be formed and the internal language system of speakers enacting those identities.
In a surprising misreading of Otheguy et al. (2015), MacSwan finds in our work a denial of the existence of distinct languages and, consequently, of bilingualism/multilingualism itself (MacSwan 2017: 168–9). But no one has denied that languages are what MacSwan calls ontological systems; the point has been all along that the dual ontology of the two separable named languages is anchored in sociocultural beliefs, not in psycholinguistic properties of the underlying system. In the ontology that we have advanced under the rubric of translanguaging, the term bilingualism and multilingualism remain as references to existing, and highly relevant, external sociocultural boundaries. But they are boundaries that, in the translanguaging conception, are not internally represented as anything other than as sociopolitical constructs (cf. the ‘translanguaging instinct’ in Li Wei 2018).
The work of García, MacSwan, Otheguy and Reid has focused heavily on English-Spanish bilingualism because of its strong presence in the United States. For us, and for MacSwan, the terms ‘English’ and ‘Spanish’ are useful for naming two sociocultural categories whose existence is uncontroversial, and whose personal and political import is considerable. But the difference is that in the familiar position regarding the competence of bilinguals, which is ultimately the one adopted by MacSwan, English and Spanish end up, in addition, as the names of two separable cognitive-structural sets of linguistic components that continue to neatly correspond to what society calls English and Spanish. The focus on English and Spanish notwithstanding, the scope of the present debate is not limited to the U.S., but is rather intended by all parties to be of universal application to all settings and languages.
Before proceeding, some conceptual and terminological clarifications are in order. In his critique of our work, MacSwan proposes what he calls a new integrative multilingual model. In the present paper we will show that, despite the term, there is little in the model that is new or multilingual. Specifically, we will argue that the model presented in MacSwan (2017) constitutes, in all essentials, a reiteration of a long-established theoretical posture that fails to grapple with recent proposals offering fresh understandings of what it means to be bilingual. This fresh understanding has gone under several names. A number of scholars have used the term polylanguaging (Jørgensen 2008; Jørgensen et al. 2011). Others have used metrolingualism (Otsuji and Pennycook 2010; Pennycook and Otsuji 2015). But, along with many other scholars, we continue to prefer translanguaging (Blackledge and Creese 2010; Blackledge and Creese 2014; Canagarajah 2011; Canagarajah 2013; Cenoz and Gorter 2015; Creese and Blackledge 2010; García and Wei 2014; García and Otheguy 2015; Gort 2015; Hornberger and Link 2012; Lewis et al. 2012a; b; Li Wei 2011; Wei 2018; Lin 2014; Otheguy et al. 2015; Sayer 2013; Swanwick 2016). In this paper as in our previous publications, we use the terms monolingual, bilingual, multilingual, and linguistic system in the ways that have become familiar. We will only make a distinction between bilingual and multilingual when necessary, and for the most part will use bilingual for both. And we will mostly use linguistic system to refer to the individual’s practice-based stock of lexical and structural features that in part make possible linguistic communication, and human interaction more broadly.
The term linguistic system thus covers for us the same territory as linguistic structure and linguistic competence. But we stress that for us system, structure, or competence is the assembly of features that, in cooperation with other resources, enables linguistic practices. That is, the linguistic system (or structure or competence) is daily enacted in, and is shaped and reshaped by, social interaction. In addition, and importantly, we think of the linguistic system (or structure or competence) as having been acquired in the first place through social interaction and communication. The linguistic system, then, is the collection of lexical and structural resources seeded by linguistic practices and grown and harvested in such practices. It is with these provisos in mind that system, structure, and competence are for us essentially equivalent terms. In our preference for linguistic system, we simply wish to underscore the distance of our position from that of Chomsky, who continues to insist that his notion of linguistic competence bears no relation to linguistic practices and is largely disconnected from human interaction and communication (Chomsky 2002: 76ff; Chomsky 2012: 12ff).
In what follows, we begin with our rebuttal of MacSwan’s claim that his patently dual proposal is not really dual. And we defend against MacSwan’s criticism of our notion of a unitary competence. We then offer three plausible arguments, drawn from multiple disciplines, that we believe deprive MacSwan’s position, and the familiar dual conception of bilingualism more generally, of any initial plausibility. Then, in the last section of the paper we offer an extensive rebuttal of MacSwan’s claim that what is usually called code switching nails down the case for a dual conception of the bilingual’s competence. We show that MacSwan’s analysis of the utterances of English-Spanish and English-Farsi bilinguals fails to meet basic requirements of observational adequacy, and that it advances, in addition, an invalid argument that reaches as conclusions only the positions that it initially assumed in its premises. Finally, we discuss the consequences for the education of language minoritized children of adopting MacSwan’s dual conception of bilingualism instead of our unitary translanguaging one.
The dual correspondence theory
We use the term dual correpondence theory to describe the widely held conception of the bilingual’s linguistic system as internally divided into two parts that correspond to the two externally named languages. This is the theory challenged in general by the translanguaging approach and specifically in Otheguy, García & Reid (2015), and it is the theory that, with some qualifications, is upheld by MacSwan (2017). To understand the dual correspondence theory as applied to, for example, English-Spanish bilinguals, consider the interpretation usually given to the statement that dog is a word of their English and perro a word of their Spanish and that, furthermore, the placement of direct object pronouns following the verb (I saw it) is a rule of their English but the placement of such objects preceding the verb (yo lo vi) is a rule of their Spanish. Under the position advanced in MacSwan (2017) these statements would make reference to the internally separate sets of words, rules, and boundaries demarcating the two distinct linguistic systems of the bilingual. In contrast, under the position advanced in Otheguy et al. (2015), the words and rules would also be part of internal linguistic system of the bilingual, but not the boundaries of the named languages nor the membership of some words and rules in one system and of other words and rules in another. The boundaries and the separate membership would be, under our translanguaging proposal, simply a matter of external, malleable and contingent convention. The reason we call the position that regards the boundaries and the separate membership as psycholinguistically real a dual correspondence theory is to underscore the theory’s proposal of cognitive duality and to stress the direct dependence of this duality on named languages and boundaries. We see MacSwan (2017) as an effective adoption, in fact a spirited defense, of this theory, succintly expressed in his insistence that we need ‘some internal language-specific differentiation’ (2017: 179, our underlining). Yet, for three reasons that require our consideration, MacSwan denies that he’s an advocate of what we’re calling dual correspondence.
First, as MacSwan sees it, he does not in any substantive way rely on the names of the two languages (in this case English and Spanish); they could be called X and Y. But even if we called them just X and Y, there would still be for MacSwan a boundary that would make X and Y cognitively separate and distinct. And what’s more important, the separate internal representations in MacSwan always match up exactly with what the society calls English and Spanish, even if he called them X and Y. Protestations to the contrary notwithstanding, MacSwan consistently uses the names of the two languages as firm guides for assigning utterances, parts of utterances, and elements of the underlying system to what for him is either one psychologically real language or the other. MacSwan (2017, page 181 among others) speaks of English words and Spanish words, English sentences and Spanish sentences and, more important, of English rules and Spanish rules, with the English or Spanish affiliation of these words and rules presented as self-evident and with no supportive argumentation. That is, MacSwan determines which of these systemic features belong to which of the two psychologically real competences solely on the basis of the two named languages. Faced for example with a U.S. Latino college student who says los teachers me hicieron flunk the class, MacSwan would be able to tell that los, me, hicieron, and the preverbal placement of me are part of a psychologically real Spanish, and that teachers, flunk, the class, and the postverbal placement of the class are part of a psychologically real English, and he would do so on no basis other than the social assignment of those features to those named languages. We will return to this problem, and will give more examples, in the section on code switching below.
Second, MacSwan maintains that he has moved from the conventional position claiming psychologically real duality to a new position claiming psychologically real multiplicity (thus he calls his model multilingual). But for the purpose at hand, this is a distinction without a difference. The term multilingual in MacSwan is not a reference to multiple languages in the familiar sense of the term, but a reference to the obvious fact that linguistic competence is in all speakers made up of multiple components (multiple phonemes, words and morphemes, of course, but also, depending on the theory, multiple rules, operations, constraints, schemas, grammatical systems, etc.). But these several components in MacSwan amount to a restatement of the dual correpondence theory, since the components eventually get sorted out into two superordinate cognitive categories that always correspond to one named language or the other. For the debate about the linguistic system of bilinguals, there is no difference between the claim of duality and MacSwan’s claim of multiplicity; both uncritically impute the two social categories, the two named languages, to the mind-brain of the bilingual.
Since it deals with the name and central holding of MacSwan’s model, this point is worth restating from a different angle. The debate about the nature of the linguistic system of bilinguals is not furthered when the many components that speakers know and draw on are called, most unhelpfully, ‘languages’, and when this piece of terminological legerdemain is then used as justification for speaking of a ‘multilingual’ model. The debate, to repeat, is not about whether the underlying grammar has one or many elements. It is about whether the many lexical and structural features mastered by the bilingual stand in a corresponding relationship to the two well-established social categories that the society sanctions through its two language names.
Thirdly, MacSwan also denies that his proposal is one of dual competence because he explicitly recognizes that the two underlying systems contain shared areas and points of contact. But acknowledging that the separation between two groups of components is not complete does not negate the basic separation between them. Recognizing that some areas of grammar and vocabulary are shared does not distinguish MacSwan’s proposal from the claims traditionally made by proponents of the dual correspondence theory. As we saw above, MacSwan emphasizes the need to characterize the competence of the bilingual in terms of internal, language-specific differentiation, a differentiation that always runs along the lines of the two named languages. And as we show below, this language specific differentiation is an essential part of his analysis, even if it allows for some overlap.
MacSwan’s critique of translanguaging’s unitary view of the linguistic system
In Otheguy et al. (2015) we explain that the competence of people that are said to be bilingual is unitary, in the sense that the lexical and structural components socially assigned to different named languages constitute a single aggregation of linguistic resources. MacSwan has misread our use of the term unitary as embodying the outlandish suggestion that we see bilingual competence as consisting of a single component. MacSwan states the obvious by insisting that linguistic competence is internally differentiated, constituted by distinct elements that are themselves subdivided into multiple sub-systems, whose output may even in some cases appear contradictory, as in the examples cited from Roeper (2011), Roper (2016)). All of this is well known. It unnecessarily muddies the waters for MacSwan to write that ‘the bilingual grammar makes use of structured and internally organized differentiation of some kind’ (MacSwan 2017: 181) while asserting on the same page that the grammar will have to attribute different, language specific properties to Spanish and English. These two statements make two very different claims. The former statement, that there is some kind of internal differentiation, is obvious and uncontroversial; the latter, that the differentiation is language specific and corresponds to the social division between English and Spanish, is precisely the conventional wisdom on dual representation that the translanguaging approach argues against. By repeating these two different claims as if they were the same, MacSwan makes it appear as if agreeing with the need for internal differentiation requires agreeing with language specific differentiation. MacSwan is not offering an argument against the unitary conception of the competence of bilinguals sponsored by translanguaging when he points out that this competence, as that of monolinguals, is made up of many phonological, lexical, and structural elements that are each distributed over many different structural sub-systems; he’s just stating an obvious point that is irrelevant to the debate.
The confusion is further compounded when MacSwan reminds us about the existence of multiple styles or registers. No one doubts that all speakers rely on different combinations of lexical and structural resources to engage in linguistic practices that vary with context, setting, and interlocutor, or that they deploy, ancillary to their vast and complex linguistic knowledge, a subtle and nuanced metalinguistic knowledge that tells them when it is appropriate to say what to whom. This is all true but irrelevant to the debate, which is about whether or not the complex and highly differentiated form that the linguistic system of bilinguals ultimately maps onto the dual social invention embodied in the names of the two languages.
Multidisciplinary arguments and evidence against the dual correspondence of the social and the biocognitive
Before proceeding further, it will be useful to introduce into the discussion three considerations that deprive the dual correspondence theory, widely held and reaffirmed by MacSwan, of the most basic initial plausibility. First, and in the most general terms, one seldom finds in any aspect of human nature a direct ontological correspondence between social and biological categories of the kind that is ultimately involved in the conventional wisdom that MacSwan aims to sustain. For example, in many societies there are racial terms that offer a socially established categorization of people. In the United States, the racial categorization rests on such terms as African-American, Asian, black, brown, Hispanic, Latino, Native American, Pacific Islander, white, etc. Such terms, in the U.S. and elsewhere, display the fluid and contested character typical of sociocultural affiliations; they change with time, they differ within subcultures, and differ again when used by different people in different settings even within the same subculture. For example, in the U.S., immigrants from Latin America who were regarded as blanco/a in their countries of origin often discover that they’re not regarded as white in the U.S., or that they are so regarded by some people but not by others. In many families in the U.S., some members think of themselves as Latinos or Latinas while siblings of very similar appearance think of themselves as African-American or white. In some U.S. families, minority racial identities are embraced, while in others they are not (often leading to people being described as, or accused of, being someone who is trying to ‘pass’). Moreover, it commonly happens in the U.S. that some people adopt one racial identity at an early point in life and a different one when they’re older.
With these disagreements and changes in mind, it would be highly implausible to propose that the current U.S. races find coterminous counterparts in biological categories of the phenotype or in the make-up of the genotype. Phenotype and DNA remain the same in the individual who traveled from Latin America and discovered that his or her race had changed. They remain the same too in the person who changed identities during the course of a lifetime. And they remain the same in siblings and other family members who argue about their racial identities.
This fluidity of racial naming is paralleled by the indeterminacy of language naming found in the anthropological present, including, importantly, situations of no naming at all. The ancient forerunners of present-day Vietnamese had no name for their languages before the start of Chinese domination in antiquity, when the imperialists coined the term Viet Nam, ‘Southern Barbarians’ from which Vietnamese is derived (Su et al. 2004). As we have known since the pioneering work by John Gumperz (1976), many people in pre-colonial Africa and India engaged in linguistic practices undergirded by idiolects that challenged the preconceptions of Western authorities, missionaries, and scholars, whose insistence on grouping these idiolects into named languages did considerable violence to the facts on the ground (Canagarajah 2013; Khubchandani 1997; Makoni and Mashiri 2007; Makoni and Pennycook 2007). Even squarely within the more familiar Indo-European area, people today continue to argue about the boundaries and names of their languages. Hindi vs. Urdu, Catalan vs. Valencian, Serbian vs. Croatian, Spanish vs. Spanglish are but a few examples of situations where the ‘one language or two’ issue is impossible to settle by lexical and structural considerations alone. The arguing is to be expected because we are dealing with contestable social categories, and would be unexpected if one were really arguing about categories of biology and the mind. The representation of the linguistic competence of the inhabitants of Viet Nam could not have been different before and after the Chinese arrived and asserted the existence of the Vietnamese language; the linguistic system undergirding successful use of language by Africans and Indians before the coming of the Europeans almost certainly did not change when the new names took root; the nature of linguistic practices and the underlying linguistic system is not likely to be different in those who believe that they speak two languages, Catalan and Valencian, and those who believe that they speak only one.
It is not surprising that one finds near universal scholarly agreement about the non-biological, historically-contingent and socially-conditioned status of racial terminology. This is the same status that we advocate for linguistic terminology, which is no less historically-contingent and perhaps even more socially conditioned. (Were it otherwise, Ethnologue would not have to caution its users that the 7,000 languages listed in their inventory coexist in the anthropological present with over 40,000 language names.) Romaine (2000) is on the mark when she asserts that ‘the very concept of discrete languages is proably a European cultural artifact fostered by procedures such as literacy and standardization’ (p. 12), and Khubchandani (1997) is no less so when he describes the essense of Indian plurality in terms of ‘fuzziness of language boundaries’ and ‘fluidity in language identity’ (p. 87). The implausibility of the dual correspondence model of the linguistic system of bilinguals whose two parts match up with the two named languages is perhaps best captured by Makoni & Pennycook: ‘At the heart of the problem here is the underlying ideology of countability and singularity, reinforced by assumptions of a singular, essentialized language object …’ (Makoni and Pennycook 2007: 10).
Racial terms are of crucial importance and of enormous consequence for individuals and communities, and indispensable for advocacy and the maintenance of efforts toward social justice. But still, they are not the names of facts represented in the body of the person whom the society describes as belonging to a race. In parallel, linguistic terms too are of crucial importance for individuals and communities, and indispensable for advocacy efforts and fomenting the maintenance and revival of the languages of linguistic minorities worldwide. But like racial terms, named languages establish soft and contestable cultural boundaries that cannot be assumed to stand in correspondence to hard cognitive boundaries that represent different language systems. In other words, just as we would reject as unsustainable the claim that the boundaries between named races map neatly onto the body (that they correspond to biologically represented boundaries in the genotype), we should reject the claim that named languages are susceptible to a similar mapping (that they correspond to biologically represented boundaries in the mind-brain).
A second consideration contributes to reducing the initial plausibility of the widely held dual correspondence theory endorsed by MacSwan. When experimental subjects in psycholinguistic studies are instructed to use only one of their named languages, the dual correspondence theory should lead us to expect that one of the two separate (sets of) competences would be activated, and that the other one would be set aside. But in a robust set of findings, what happens is exactly the opposite. As Kroll and Bialystok (2013) put it, ‘the key discovery in the research on bilingualism is the overwhelming evidence … that both languages are active to some degree when bilinguals are using one of them’ (Kroll and Bialystok 2013: 498). Which is to say that the bilingual does not behave in experimental settings as someone operating with a dual competence that leads to the activation now of one language, now of the other.
With reference to lexical items, there is considerable specific evidence that bilinguals possess what is essentially a single holistic network, a single lexicon that is not partitioned along the lines of the named languages (Costa 2005; Dijkstra and Van Heuven 2002). The familiar statement found in psycholinguistic research expressing the conclusion that ‘both languages are activated’ at all times is, of course, the inevitable result of an initial assumption of duality on the part of the researcher. For to find that the two languages are always activated is to weaken considerably the claim that there are two separate languages to begin with. In short, the key finding of psycholinguistic research on bilingualism does not at all support the notion that the bilingual has a binary competence and a binary system, and in fact tilts initial plausibility in the direction of the translanguaging claim of a unitary linguistic system.
A third consideration militating against the conventional view of dual competence in bilinguals sustained by MacSwan has to do with the treatment of what are usually called loanwords. In an extensive corpus-based study, Poplack (2018) finds that bilinguals tend to treat loanwords from the moment of first borrowing according to the structural patterns of the native words of the receiving language. Relying on more than ten different situations of bilingualism, including that involving French and English in Canada, Poplack shows that the morphosyntactic makeup of borrowed items, and their statistical distribution with respect to morphology and syntax, is the same as that of native items from the start. Adaptation to the grammar of the recipient language is abrupt, not gradual; well established loanwords did not start out as one-word code switches that preserved the morphosyntax of the donor language and gradually became nativized.
Although Poplack does not weigh in on the debate about the nature of the competence of bilinguals, her findings suggest, as did the experimental findings mentioned above, that the difference between native and foreign elements of competence is not structural but social. We interpret Poplack’s findings as supporting the following understanding of U.S. Latino speech on the occasions when they are speaking in what the society calls Spanish and are using words in the plural like los biles ‘the bills’. These speakers never went through a period where they said los bills, with a lax vowel [I] and a dark consonant [L], and especially never pluralized the word with a voiced suffix [-z]. Rather, in this community the word was, from the start, formed with tense vowel [i] and light consonant [l], and with a plural consisting of a vowel and a voiceless fricative, [-Es]. There is thus not now, and there never was, anything in this noun structurally different from any other noun used in this community. That is, there never was anything in this process of lexical expansion that would justify invoking ‘English’ and ‘Spanish’ as linguistic cognitive categories. Our knowledge that cuenta ‘bill’ is a Spanish word and that bill is a loanword is a piece of sociocultural knowledge. In linguistic terms, borrowed and native words are the same, both residing in an undifferentiated lexicon. Again here, little solace is to be found for the notion of two linguistic systems and a dual competence underlying the behavior of bilinguals.
Linguistic theory and the debate about the linguistic system of bilinguals
In a fully developed debate about the nature of linguistic system in bilinguals, the systems and sub-systems that together make up this competence would naturally need to be articulated through specific theoretical constructs belonging to a specific theory of language. The identity of these constructs would be entirely dependent on the choices the analyst would make regarding the many substantively different theories of grammar that guide the study of language today. Participants in the debate would adopt theories that incorporate many, but almost certainly not all, of the following: a lexicon, a phonology, a morphosyntax, a collection of constructions, a set of meaningful structural schemas, a set of meaning-carrying grammatical systems, a computational system, a sentence, a clause, a lexical categorization, a pronominal system, a tense system, an autonomous syntax, a movement rule, a device of lexical insertion, a statement of strict subcategorization, a projection of syntax from lexical entries, a logical component, to name only a few. (To insist, these elements would not all be found in the same analyst’s theory; they are a list of what one is likely to find in different theories.) But even though we will have to recognize at some future stage of this discussion that the choice of grammatical theory is likely to influence the way we think about the linguistic system of bilinguals, it bears stressing that the rejection or acceptance of a particular grammatical theory and its constructs has not so far figured explicitly as a factor in the debate.
It is of course true that MacSwan (2017) couches his points in terms of the constructs of a Chomskyan generative grammar. Moreover, the tone of his writing is likely to induce in some readers a belief in the endearing fiction that linguistic theory begins and ends with the assumptions, tenets, and intellectual tendencies of generative theory. And it is also true that MacSwan’s choice of generative theory to frame the debate is a curious and perhaps unfortunate one. (For while the debate is about the capacity that underlies the bilingual’s natural, real-time use of language, generative theory stands out for relying predominantly on introspective data, and for a stance that avowedly disconnects it from, and makes it only very indirectly relevant to, the study of language use [Chomsky 2002: 76; Chomsky 2012: 97, 167]). But whether the choice of formal generative theory is felicitous or not, the point is that the well-known dual correspondence position adopted by MacSwan is not dependent on formal generative grammar, nor has it been shown by him to be so dependent. Many linguists following many other theoretical approaches have for years believed that the two named languages of the bilingual correspond to separate and discrete underlying systems and representations.
We of course welcome MacSwan’s reassertion of the view that an underlying linguistic system is needed to explain human linguistic behavior, even as it is shaped by social practice and experience. Sociolinguists working with translanguaging often limit their discussion to linguistic practices and seldom or never mention, or explicitly deny, the underlying grammar that enables practices even as it is shaped by them. We proceed, in agreement with MacSwan, under the assumption that there is something called system or competence whose nature in bilinguals it is possible to disagree about. Our observation is simply that, irrespective of one’s preferred linguistic theory, no actual evidence has so far been offered that bilinguals have two discrete systems or groups of systems that correspond to the two named languages that the society attributes to the bilingual.
In Otheguy et al. (2015), we proposed that another way to make our point about unitary competence is to say that the idiolects of bilinguals, their own individual linguistic competence and associated linguistic practices, are qualitatively like those of monolinguals, in that, like the monolinguals’, they’re unitary rather than being carved out in separable compartments corresponding to the two named languages. In our view, monolinguals and bilinguals differ only quantitatively; the idiolect of the bilingual contains more lexical or structural elements, but these, when taken together, make up a single aggregation of elements.
MacSwan (2017) sets aside our idiolect, preferring instead Chomsky’s notion of the ‘I-language’ (i.e., Internalized language). I-language is the linguistic knowledge of an individual speaker and is distinct from the externalized manifestations of language, which Chomsky calls E-language (i.e., Externalized-language). E-language includes both language use and the sociolinguitically defined categories of named languages. Granted, Chomsky’ I-language may very well by largely equivalent to our term idiolect; but we find the term idiolect preferable because it is not connected with any particular linguistic theory; and as we have seen, there is little indication so far that the present debate hinges on the adoption of one or another linguistic theory, making the more theory-neutral term preferable. The term idiolect is also preferable to I-languge because it captures in its very form (cf. ‘idio-’) the fact that many details of the linguistic sytem are unique to each individual. In the translanguaging view, this individual finds his or her idiolect socially yoked to similar idiolects only by the social sanctions instituted through the named language.
Finally, it is worth noting that the use of I-language by MacSwan in the context of the present debate is not benign. Despite protestations to the contrary, the generative framework effectively assumes the linguistic reality of named languages (e.g. statements such as “Spanish is a pro-drop language” occur frequently in the generative literature without apology or qualification). The generative framework has thus inadvertently incorporated named languages into its version of the mental grammar (the I-language) of monolinguals and bilinguals, assuming for the latter, from the start, a dual correspondence theory.
Three elements of generative analytical practice reflect this assumption. First, the judgment that an utterance is grammatical or ungrammatical is meaningless without specifying the linguistic system in terms of which it is or isn’t grammatical, and named languages are the obvious candidates (e.g. yo lo vi with a preposed object pronoun is grammatical in Spanish but I it saw is ungrammatical in English). Second, the introspective data upon which generative analyses largely rely are produced by bilingual informants who themselves believe they speak two languages, and that belief colors all their judgments. Third, the generative framework has not developed a methodology for distinguishing introspective data reflecting features of I-language from introspective data reflecting features of E-language; for instance, there is no basis for deciding whether the fact that English-Spanish bilinguals agree that casa is a Spanish word and house is an English word reflects a feature of I-language or a feature of E-language; or whether a judgment of unacceptability reflects a fact of I-language or E-language.
The empirical argument and the significance of code switching
MacSwan builds the case for his position and for the dual correspondence theory more generally by invoking the notion of code switching and presenting it as an established fact of bilingual behavior. Again voicing the social concern that we have noted, MacSwan writes that the concept of code switching has helped to place the behavior of speakers from linguistically minoritized communities in a positive light. The alternation between codes, MacSwan reminds, has been demonstrated to be systematic, a demonstration that counters the pernicious charges of haphazardness and alinguality that have long been leveled against the speech of many bilinguals. And since code switching is most coherent within a dual correspondence conception of the bilingual’s competence (the bilingual is switching from one system, or code, to the other) it makes sense to see code switching as not only politically beneficent but as evidence for the dual nature of the bilingual’s competence.
With respect to the social issue, the obvious needs stating: the felt need to protect bilinguals from social derogation by showing the systematicity of the process of switching between languages neither adds nor detracts from the argument. Moreover, the need for protection itself comes from the very fact of positing two codes in the first place. The demonstration of systematicity in switching rests on the assumption that the dual correspondence theory of underlying bilingual competence is correct, and that what one is hearing in a bilingual family or community is indeed people switching between psycholinguistically real languages. But under what we consider to be unitary competence, there would be no code switching going on, nor any need to defend people from the charge of asystematicity.
Saying it in different words, bilingual speech is seen as switching only because many linguists, MacSwan among them, and bilinguals themselves, are social actors whose interpretation of linguistic behavior is guided by the named languages and by the idea that these named languages correspond to two linguistic systems. That is, code switching is a by-product of the dual correspondence theory. In strict theoretical terms, the haphazardness and alinguality that well-intentioned scholars want to defend the bilingual against are creatures of their own making. When, as in the position adopted under the translanguaging approach, there is only one underlying system, the behavior cannot be characterized as switching codes; and when there’s no switching it makes no sense to characterize it as unsystematic or haphazard. (It is of course a different story that many laymen will continue to regard the speech of people called bilingual as an odd mishmash; but discussion of this problem, which parallels the similar derision reserved for the speech of many people known as monolingual, would take us here too far afield.)
Turning the discussion of code switching from the political to the structural, consider first MacSwan’s treatment of the asymmetry between noun phrases consisting of determiners and nouns in which, in MacSwan’s telling, the determiner belongs to one language and the noun to the other. Citing the research literature on Spanish in the U.S., MacSwan maintains that cases of Spanish determiner with English noun, e.g. los teachers (or the los biles mentioned above) are well-formed, while cases of English determiner with Spanish noun (e.g. the casa) are ill-formed and in fact absent from corpora. MacSwan regards this as an ‘intriguing challenge’ that provides evidence for the need ‘to posit an underlying system that attributes different properties’ to Spanish and English (MacSwan 2017: 181), all supporting the need to posit, as we have seen, language-specific internal differentiation. We first discuss the empirical failure of the analysis and then turn to important theoretical problems that the analysis overlooks.
Our first critique of MacSwan’s analysis is empirical. It is simply not true that noun phrases like the casa are absent from the speech of bilingual U.S. Latinos. Anyone familiar with the community repeatedly hears speakers saying, in stretches of what is conventionally known as English, noun phrases like the quinceañera, the escritorio, the vacinilla, the cuñado, the pendejo, the caja de los juguetes (‘the sweet fifteen, the desk, the potty, the sucker, the toy box’) and many others like them. During the weeks profitably spent reading MacSwan’s paper, one of the present authors had occassion to be instructed by a bilingual Latina on the use of a baby stroller, as follows: Make sure to put your foot under this thin black bar here, and then bend down and take the palanquita that you see right there and pull it out. Here the palanquita ‘the lever’ is counter to MacSwan’s generalization. On that same day, one of the authors watched a baseball game where a large billboard advertising beer proclaimed in bold letters The light cerveza, only to return home to hear another bilingual Latina say, Hey, we’re not leaving; the pollitos that Mami makes for dinner are worth the wait. Once more, the cerveza ‘the beer’ and the pollitos ‘the chicken cutlets, lit. the little chickens’ appear to contradict MacSwan on empirical grounds. The case that MacSwan makes for what we’re calling the dual correspondence theory on the basis of the different treatments accorded to code switches between English and Spanish determiners and nouns rests on data of questionable validity.
A second set of data offered by MacSwan in favor of thinking of certain utterances by U.S. Latinos and Latinas as a type of orderly code switching that supports language-specific internal differentiation has to do with the placement of adjectives and nouns. But here again the evidence is weak. Observers of the U.S. Latino community and of the world Hispanic community in general will learn to their surprise in MacSwan (2017: 181) that ‘a Spanish English bilingual’, displaying the systematic application of ‘the English rule’ to ‘English words’ and ‘the Spanish rule’ to ‘Spanish words’, would ‘never utter’ house white or blanca casa (note, in passing, the reliance on the named language to apportion words and rules to languages). MacSwan claims that the person’s underlying system would generate the white house, la casa blanca, and the white casa, but never la blanca casa or the casa white. The argument is based on the false claim that sequences of adjectives preceding nouns are ungrammatical and never found in the language of Latinos or, as MacSwan puts it, that the Spanish rule that applies to Spanish words forbids preposed adjectives. This is another fact enlisted in support of code switching, of its orderliness, and of its constituting evidence for two separate grammars, that turns out not to be a fact at all. MacSwan’s supposedly ungrammatical blanca casa reflects the commonest of structures in the language of Hispanics world wide, for whom la blanca nieve, el bello atardecer, la fiel compañera (‘the white snow, the beautiful sunset, the faithful companion)’, to say nothing of el buen amigo ‘the good friend’, or even the greeting buenos días ‘good morning’ are but a few of tens of thousands of similar sequences of adjective preceding noun used in unmarked registers (for what it’s worth, a Google search for the supposedly ‘never uttered’ blanca casa yields 459,000 entries). The analysis of Spanish nouns and adjectives offered by MacSwan in support of his code switching argument is contradicted by the facts. A correct analysis of Spanish adjectives and nouns, couched in the generative theoretical terms which MacSwan favors, would have to say that the syntax of Spanish freely generates grammatical sequences of [A+N], to which a semantic component then assigns a different interpretation than is given to the also freely generated sequences of [N+A]. Translated into other theoretical terms, one would say that in the case under discussion, as in many other parts of the grammar, word order is meaningful. It is not that [A+N] is ungrammatical, but that it means something different from [N+A]. This fact is well known to scholars of Spanish linguistics, being a regular entry in all standard grammars (cf. Alarcos Llorach 1994: 81), and having received a rich and detailed analysis in Klein-Andreu (1983).
In trying to make sense of the adjective-noun studies reviewed by MacSwan, one can only speculate that he and the authors he cites were told by informants that they would never say la blanca casa because they found it inappropriate for the experimental contexts presented to them, for which la casa blanca probably made much more sense. And the same is most likely the case with the presumed ungrammatical code switches the house white and the casa white. As in the cases above, we think that reports of the lived linguistic experience of Latino bilinguals are relevant: During the writing of this piece, we overheard a different Latina woman say to her child ¿Querés ir al basement de la otra tienda y compramos esos cositos [sic] cool?, ‘Do you want to go to the basement of that other store and buy those little cool thingies?’ Here cositos cool is an impermissible switch according to MacSwan (because cositos would be a Spanish N followed by cool, which would be an English A, like the supposedly ungrammatical casa white). On the same day we overheard a conversation by a couple making a bed where one interlocutor said Estas sábanas no son deep, only to be answered ¿Por qué no compras sábanas deep?, where sábanas deep is again presumably not allowed by the rules that MacSwan tells us establish the systematicity of code switching and demonstrate the existence of the two underlying sets of grammatical elements in bilinguals that correspond to the two named languages.
Turning now from the empirical to the theoretical, it is worth showing that, even with better data, the analyses offering a contrast between presumably occurring noun phrases like los teachers and presumably non-occurring ones like the casa, and of occurring the white casa and non-occurring the casa white are deeply flawed. We first note again that MacSwan’s analyses start with the premise that, in real cognitive terms, there are two linguistic systems, or codes, that is, that there are English rules and words and Spanish rules and words. We note too that the named languages are used as guides to the presumptive fact that the words the, teachers, and white are English while the words los, casas, and blancas are Spanish. Given these assumptions, the sequences los teachers and the casa represent, of course, code switches. And if it were true that los teachers were well formed while the casa were not, then we would indeed have an asymmetry in the combinations of English and Spanish determiners and nouns, which would in turn support the conclusion of language specific internal differentiation. And the same reasoning would apply to English and Spanish adjectives and nouns.
But the only element in the conclusion of MacSwan’s analysis that was not already assumed to start with is the pattern of presumed well- and ill-formedness. The prior and independently necessary demonstration that, for example, the speaker who reports the well-formedness of los teachers and the ill-formedness of the casa had extracted teachers and casa from two psycholinguistically separate lexicons has not been made. The outcome of the analyses, even if they were empirically sustainable, was pre-ordained by premising the duality of membership of the lexical and grammatical forms involved.
This criticism of MacSwan’s position bears repeating in different terms. The point in need of demonstration is the existence of switching between elements of two discrete underlying systems. But the analysis turns this point in need of demonstration into an assumption. Guided by their membership in the two different named languages, MacSwan has assumed that casa and teachers belong to two different cognitive representations, two different linguistic systems, and that the same is true for blanca and white and los and the. On this basis, he has assumed that these utterances are code switches, not just in the eyes of the society but in psycholinguistically real terms. By assuming that which is in need of demonstration, MacSwan has not established that the reason for the pattern of speaker judgments is the combinatorial asymmetry of psychologically real linguistic units of heterolinguistic provenance; and he has not established, of course, the cognitive separability of the two linguistic systems corresponding to the two named languages.
Another deep theoretical flaw in MacSwan’s discussion of what he calls code switching is that the facile assignment of words to languages ignores the widespread phenomenon of lexical borrowing. Even though the term loanword has its origins in studies of language change, it is now widely accepted that loanwords represent also a widespread synchronic phenomenon. In the standard telling of the dual correspondence theory, it is the common experience of bilingual communities that words from what the society considers one language become incorporated into the lexicon of another. This means that when Latinos in the U.S. are said to be speaking in Spanish, and yet use words like el bil, el sobbuey, la boila, el bildin, el apoinmen, el jáiscul or any of the others of the los teachers type, they may not be using at all the English words bill, subway, boiler, building, appointment, or high school, but using instead what for them are new words of Spanish. Similarly, and more important for MacSwan’s point, the common fact of widespread borrowing means that items of the the casa type used by the same speakers (like the already mentioned the quinceañera, the escritorio, the vacinilla, the cuñado, the pendejo, the caja de los juguetes, the palanquita, the pollito, etc.) may not contain for these speakers any Spanish words, but what for them are English words. How does MacSwan know, then, that the speaker who produces, or assents in an experiment, to los teachers but fails to produce, or rejects in an experiment, the casa is code switching? If casa is an English word, the casa is not any more a switch than the house, a fact that is equally true of the patio, the junta, or the guerrilla, the latter being all words that everyone now is likely to regard as English even though they were all once regarded as Spanish.
In methodological terms, the phenomenon of lexical borrowing makes MacSwan’s analysis unfalsifiable. If, faced with our counterexamples, MacSwan were to have the perfectly plausible reaction that cerveza, quinceañera, palanquita, vacinilla, pollito etc. are not counterexamples to his analysis because they are recent loanwords residing in the speaker’s newly expanded English lexicon, the analysis would in principle lack any possible falsifying evidence. Each potential counterexample to the claim that the construction [English Det+Spanish N] is ungrammatical would be ruled out as evidence by turning it, through borrowing, into the grammatical [English Det+English N].
The general flaw of using the named languages as guides for assigning certain stretches of bilingual discourse to one language, and other stretches to another, is one that has been pointed out before, and that MacSwan denies outright. Code switching research, says MacSwan, is not conducted on the basis of an a-priori tagging of discourse fragments as belonging to one or the other language. Code switching researchers ‘have taken care to conceptualize language differences according to the structural patterns evidenced by speakers, not according to the political identification of languages’ (MacSwan 2017: 182). But as we have seen, these statements remain at the level of unfounded assertion.
Our criticism of the assumption that words and rules can be assigned to languages in disregard of the problem of borrowing should be articulated in more general terms. To our knowledge, the required exercise has not been carried out starting from a clean slate, that is, studying the utterances of bilinguals in disregard of the social dictates that tell the researcher which stretches belong to one language and which to the other. A thought experiment makes the point: Imagine a transcribed corpus of speech from a thoroughly unstudied part of the world where the prevailing culture is mostly unknown to us. All we know about the speakers who produced the corpus are three things in which they happen to resemble us: They have a cultural notion of ‘language’ similar to ours; they see themselves as bilingual; and they say that they often do what we would call code switching or, in their terms, going back and forth from one language to the other (we will call them language A and language B because the speakers haven’t provided us any language names). The corpus comes untagged and offers the linguist no social information of any kind; we don’t know whether it is a corpus of language A or of language B, nor do we know whether this particular corpus contains code switches, nor, if it does, where in the transcript are the utterances in which code switching has taken place and where the ones where it hasn’t. MacSwan maintains that code switching researchers would be able to tell us which parts of the corpus belong to A, which to B, and which represent code switches and which do not. We look forward to being pointed to such a demonstration. And we would be especially interested to know (and especially keeping in mind Poplack’s findings discussed above) what kind of structural evidence the researchers would be looking for to spot an instance of code switching. For now, and on the evidence presented by MacSwan, of which the presumably code switched los teachers and the casa are but one example, we see that in fact, and at every turn, the identification of items as belonging to one or the other language is assertively assumed entirely on the basis of the conventional identification of fragments under the guidance of the two named languages.
In discussing what society knows as English-Farsi bilinguals (whom we will discuss only briefly, as we do not know their linguistic system well enough to comment), MacSwan shows a pattern that he has also pointed out for what society calls English-Spanish bilinguals. In MacSwan’s telling, a Farsi verb, such as dad-e, always appears with a preposed object, whereas an English verb, such as give, appears with a postposed one. To be sure, this is evidence that different lexical verbs (or sets of lexical verbs) in the bilingual’s linguistic system project a different syntax. But this does not provide evidence for the existence of two separate grammars; all it does is tell us just that, that different lexical verbs are associated with a different syntax. Linguists have always known that in all speakers, whether they’re socially regarded as monolingual or bilingual, the details of syntax often need anchoring on specific lexical items. There is little to be learned from the fact that the syntax projected by give is different from that projected by dad-e, since it is also different from the syntax of push and of rain and of many other verbs in the single lexicon of the bilingual. These familiar specializations of syntactic patterns to specific lexical items do not add much at all to the debate; unless one already knows, from information provided by the society, which lexical items belong to which of the proposed separate systems.
Educational implications of adhering to the dual correspondence theory
Joining other scholars working on translanguaging (see references above), Otheguy et al. (2015) sketched out arguments for the need to transform instruction and assessment based on translanguaging theory. MacSwan’s (2017) critique of our work focused mostly on linguistic arguments that we have shown to be invalid, but brought up as well some implications for the welfare of minoritized bilingual and multilingual communities, in particular these communities’ students. MacSwan maintains that our unitary theory undermines multilingualism, whereas his dual correspondence theory supports it. On the contrary, the unitary theory of translanguaging offers a linguistic understanding of bilingualism and multilingualism that has large implications for the just education of bilingual and multilingual minoritized children.
We’ve pointed out above that our unitary conception of linguistic competence in bilinguals and multilinguals does not negate the existence of bilingualism or multilingualism, the importance of which in the lives of members of minority communities we have ever amply supported and continue to assiduously encourage. We have merely insisted on recognizing the duality of bilingualism (or the multiplicity of multilingualism) as a product of socio-political categorization rather than a description of psycholinguistic reality. And we have insisted, based on our extensive classroom experience, that when socio-political categorization is mistakenly assumed to correspond to psycholinguistic reality, dynamic translanguaging practices and the panoply of semiotic resources deployed by bilinguals and multilinguals have very often been ignored or, worse, misunderstood and devalued, resulting in serious problems of stigmatization and exclusion.
The problems arise because the dual correspondence theory leads educators to conceive of the named languages of emergent bilinguals as linguistic objects, when in fact the only linguistic objects are the phonological, lexical, structural, and other semiotic resources at their disposal. All the recent educational standards and the assessments that measure them are based on the strong belief that English is a linguistic object, as are Arabic, Chinese, French, German, Spanish, Swahili, etc. This conception leads to an education that stigmatizes bilingual speakers whose practices differ from those that schools regard as valid manifestation of these incorrectly conceived linguistic objects. To take the most obvious case as illustration, student failures to distinguish sharply between languages, or the persistent ‘mixing’ of, or ‘switching’ between, them is not seen by the educator as a failure of social convention, of aligning one’s behavior with a social norm, but as a true shortcoming involving the mishandling of mental boundaries seen as cognitive objects. In the case at hand, the contention that English and Spanish must be seen as possessing, in MacSwan’s terms, language specific internal differentiation affirms that they are distinct linguistic objects that can be taught and assessed separately. This supports the expectation of many educators, long made suspect by Grosjean (1982), that bilinguals can be profitably understood as two monolinguals in one, and that teaching and testing should proceed accordingly.
The dual correspondence theory of the bilingual’s linguistic cognition bears responsibility for much miseducation of minoritized speakers. Whether these students are in English as a second language or bilingual programs, they are expected to act as if they should hold their two named languages as separate cognitive-linguistic entities. When bilingual students persist in displaying their translanguaging abilities, educators continue to regard these practices as truly cognitive ‘errors,’ or ‘interference’, and in identifying them as real symptons of psycholinguistic deficiency. In our extensive experience in English as a second language and English-Spanish bilingual classrooms in the United States, the discourse of teachers, even bilingual teachers, continues to trade in the currency of the alingual student who belongs to the group of the nilingües (ni inglés, ni español) ‘the neither language ones’, as some bilingual teachers recently told us, expecting us to partake of the dual correspondence theory that enables the misguided thinking that leads to this disdainful discourse. It is a discourse, we note, amply validated by the research community, where, in more elegant terms, the alingual and the nilingüe are cases of ‘incomplete acquisition’, another notion spawned by the misguided thinking enabled by the dual correspondence theory (for references and a critique of incomplete acquisition, see Otheguy 2016).
Nowhere are the negative consequences of dual correspondence more evident than in assessment. Bilingual minoritized students are assessed with instruments that measure attainment separately in the two real linguistic objects that the named languages are incorrectly theorized to be. Seldom are students assessed in their overall linguistic abilities or in their ability to use their full set of phonological, lexical, and structural resources for interactive and academic purposes. The prevailing measurement model, a direct manifestation of the dual correspondence theory that prevails in education circles, only values positively those test takers who manage to operate with less than half of their linguistic resources, castigating students when translanguaging practices make their presence felt in test performances.
As is true of many, perhaps most, bilinguals worldwide, for many of us in the U.S. Latino community it is of the utmost importance to sustain our bilingualism for the present and that of our children and grandchildren for the future. But to sustain bilingual futures, educators must learn to distinguish between constructed sociocultural concepts, such as named languages, from the lexical and structural features that shape the translanguaging performances of bilinguals and multilinguals. A similar distinction needs to be drawn around the notion of additive bilingualism. Educators and scholars must distinguish between, on the one hand, additive bilingualism seen as the addition of lexical and structural features to the bilingual’s expanding unitary language system from, on the other hand, additive bilingualism seen as the addition of an autonomous and separate named second language (Cummins 2017). The former is the true driver of bilingual proficiency; the latter has often been its retarding obstacle. A true additive approach should require teachers to look at their students from the start as the successful acquirers of the expanding lexical and structural resources of their unitary competence, not as the only very partially successful or failed acquirers of a second named language. The teacher inspired by the unitary conception of translanguaging presides over an always successful additive process of expansion; the teacher inspired by dual correspondence is daily witness to the not-there-yet shortcomings of children and adolescents who still have not mastered the second named language. Asking the reader’s indulgence for the play on words, we see the linguistic future in linguistic features, not in linguistic names. The educational future of our children is in the ways that they language to make meaning and to learn, not just in mastering socio-culturally named constructs and boundaries.
Educators and scholars interested in the wellbeing of bilingual students, especially bilingual children, learn to leverage translanguaging in instruction; they learn, that is, to validate the many different linguistic performances of bilingual children as ways of making meaning, of learning, of creating, of being critical, of imagining. They learn that to make meaning bilinguals have at their disposal a large, full, unitary linguistic system that is not partitioned into the linguistic compartments that schools create. And they learn to evaluate bilingual children for what they know how to do, independently of the culturally dictated affiliation of the linguistic and semiotic features they use. The translanguaging theory that we espouse ensures that educators put bilingual children first. These educators give students the affordances to expand what is truly psycholinguistically real, namely their lexical, structural, and other semiotic resources, engaging, to be sure, the importance of named languages and boundaries for schools and societies, but keeping in mind that these are sociocultural constructions without cognitive linguistic reality.
Alarcos Llorach, Emilio. 1994. Gramática de la lengua española. Madrid: Editorial Espasa Calpe.
Blackledge, A. & A. Creese. 2010. Multilingualism: A Critical perspective. London: Continuum.
Blackledge, A. & A. Creese (eds.), 2014. Heteroglossia as practice and pedagogy. New York: Springer.
Canagarajah, S. 2011. Translanguaging in the classroom: Emerging issues for research and pedagogy. Applied Linguistics Review 2. 1–27.
Canagarajah, S. 2013. Translingual practice: Global Englishes and cosmopolitan relations. London: Routledge.
Cenoz, J. & D. Gorter (eds.), 2015. Multilingual education: Between language learning and translanguaging. Cambridge: Cambridge University Press.
Chomsky, Noam. 2002. On nature and language. Adriana Belletti & Luigi Rizzi (eds.), Cambridge: Cambridge University Press.
Chomsky, Noam. 2012. The science of language: Interviews with James McGilvray. New York: Cambridge University Press.
Costa, A. 2005. Lexical access in bilingual production. In Judith Kroll & A. M. B. De Groot (eds.), Handbook of bilingualism: Psycholinguistic approaches., 308–325. New York: Oxford University Press.
Creese, A. & A. Blackledge. 2010. Translanguaging in the bilingual classroom: A Pedagogy for learning and teaching? The Modern Language Journal 94. 103–115.
Cummins, J. 2017. Teaching minoritized students: Are additive approaches legitimate? Harvard Educational Review 87. 404–425.
Dijkstra, T. & W. J. B. Van Heuven. 2002. The architecture of the bilingual word recognition system: From identification to decision. Bilingualism: Language and Cognition 5. 175–197.
García, O., S. Johnson & K. Seltzer. 2017. The Translanguaging classroom. Leveraging student bilingualism for learning. Philadelphia: Caslon.
García, O. & T. Kleyn (eds.), 2016. Translanguaging with multilingual students. New York and London: Routledge.
García, O. & Li Wei. 2014. Translanguaging: Language, bilingualism and education. London, United Kingdom: Palgrave Macmillan Pivot.
García, Ofelia & Ricardo Otheguy. 2015. Spanish and Hispanic bilingualism. In Manel Lacorte (ed.), The Routledge Handbook of Hispanic Applied Linguistics, 639–658. New York: Routledge.
Gort, M. 2015. Transforming literacy learning and teaching through translanguaging and other typical practices associated with “doing being bilingual.”. International Multilingual Research Journal 9. 1–6. doi:
Gort, M. & R. Pontier. 2013. Exploring bilingual pedagogies in dual language preschool classrooms. Language and Education 27. 223–245.
Gort, M. & S.F. Sembiante. 2015. Navigating hybridized language learning spaces through translanguaging pedagogy. International Multilingual Research Journal 9. 7–25.
Grosjean, F. 1982. Life with two languages. Cambridge, MA: Harvard University Press.
Gumperz, John. 1976. The sociolinguistic significance of conversational code-switching. Working Papers of the Language Behavior Research Laboratory 46. Berkeley: University of California.
Hornberger, N. H. & H. Link. 2012. Translanguaging and transnational literacies in multilingual classrooms: A Bilingual Lens. International Journal of Bilingual Education and Bilingualism 15. 261–278.
Jørgensen, J. N. 2008. Polylingual languaging around and among children and adolescents. International Journal of Multilingualism 5. 161–176. doi:
Jørgensen, J. N., M. S. Karrebæk, L. M. Madsen & J. S. Møller. 2011. Polylanguaging in superdiversity. Diversities 13. 24–37.
Khubchandani, L. M. 1997. Language policy and education in the Indian subcontinent. In R. Wodak & D. Corson (eds.), Encyclopedia of Language and Education, 179–187. Dordrecht: Springer Publishing. doi:
Klein-Andreu, Flora. 1983. Grammar in style: Spanish adjective placement. In F. Klein-Andreu (ed.), Discourse perspectives on syntax, 143–179. New York: Academic Press.
Kroll, Judith & Ellen Bialystok. 2013. Understanding the consequences of bilingualism for language processing and cognition. Journal of Cognitive Psychology 25. 497–514.
Lewis, G., B. Jones & C. Baker. 2012a. Translanguaging: Developing its Conceptualisation and Contextualisation. Educational Research and Evaluation 18. 655–670.
Lin, Angel. 2014. Hip Hop Heteroglossia as practice, pleasure, and public pedagogy: Translanguaging in the lyrical poetics of “24 Herbs” in Hong Kong. In Angela Creese & Adrian Blackledge (eds.), Heteroglossia as practice and pedagogy, 119–136. London: Routledge.
Makoni, S. & A. Pennycook. 2007. Disinventing and reconstituting languages. Clevedon, UK: Multilingual Matters.
Makoni, Sinfree & Pedzisai Mashiri. 2007. Critical historiography: Does language planning in Africa need a construct of language as part of its theoretical apparatus? In Sinfree Makoni & Alastair Pennycook (eds.), Disinventing and reconstituting languages, 62–89. Clevedon: Multilingual Matters, Ltd.
Mazak, C.M. & K.S. Carroll (eds.), 2017. Translanguaging in higher education: Beyond monolingual ideologies. Bristol, UK: Multilingual Matters.
Otheguy, R., O. García & W. Reid. 2015. Clarifying translanguaging and deconstructing named languages: A perspective from linguistics. Applied Linguistics Review 6. 281–307. doi:
Otheguy, Ricardo. 2016. The linguistic competence of second-generation bilinguals: A critique of ‘incomplete acquisition.’. In Christina Tortora, Marcel Den Dikken, Ignacio L. Montoya & Teresa O’Neill (eds.), Romance Linguistics 2013. Selected papers from the 43rd Linguistic Symposium on Romance Languages (LSRL), New York, 17-19 April, 2013, 301–320. Philadelphia: John Benjamins Publishing Company. [RLLT 9].
Otsuji, E. & A. Pennycook. 2010. Metrolingualism: Fixity, fluidity and language in flux. International Journal of Multilingualism 7. 240–254.
Paulsrud, B.A., J. Rosén, B. Straszer & A. Wedin (eds.), 2017. New perspectives on translanguaging and education. Bristol, UK: Multilingual Matters.
Pennycook, A. & E. Otsuji. 2015. Metrolingualism. Language in the city. Oxford: Routledge.
Poplack, Shana. 2018. Borrowing. Loanwords in the speech community and in the grammar. New York: Oxford University Press.
Roeper, T. W. 2011. Strict interfaces and three kinds of multiple grammars. In E. Rinkle & T. Kupisch (eds.), The development of grammar. Language acquisition and diachronic change, 205–228. Philadelphia, PA: John Benjamin Publishing Co.
Romaine, Suzanne. 2000. Language in society: An Introduction to sociolinguistics. Oxford: Oxford University Press (Second edition).
Roper, T. 2016. Multiple grammars and the logic of learnability in second language acquisition. Frontiers in Psychology 7. 4. doi:
Sayer, P. 2013. Translanguaging, texmex, and bilingual pedagogy: Emergent bilinguals learning through the vernacular. TESOL Quarterly 47. 63–88.
Su, Nguyen Khac, et al. 2004. Northern Vietname from the Neolithic to the Han period. In Ian Glover & Peter Bellwood (eds.), Southeast Asia, from prehistory to history, 177–208. Oxford: RoutledgeCurzon.
Swanwick, R. 2016. Scaffolding learning through classroom talk: The role of translanguaging. In M. Marschark & P.E. Spencer (eds.), The Oxford Handbook of Deaf Studies in Language, Oxford, UK: Oxford University Press.
Wei, Li. 2011. Moment Analysis and translanguaging space: Discursive construction of identities by multilingual Chinese youth in Britain. Journal of Pragmatics 43. 1222–1235.
Wei, Li. 2018. Translanguaging as a practical theory of language. Applied Linguistics 39. 9–30.