Code Switching Although it Sounds

Code Switching

Although it sounds like something straight from a James Bond action thriller, the term "code switching" is actually used to describe those cases wherein a bilingual speaks to another bilingual with the same linguistic background and changes from one language to another in the course of conversation (Wei, 2000). Code switching is also used because it can help convey more meaning than a single language provides rather than the speakers' specific preferences for the one or the other language (Veronosi, 2000). In an increasingly multicultural world, understanding how people acquire language and apply it in multi-lingual situations has assumed greater importance than ever, and, not surprisingly, linguists have contributed to a growing body of research into this area in recent years. The purpose of this paper is to provide an overview of the grammar of code switching, some constraints associated with the concept, a description of the quantitative and qualitative differences that affect language acquisition and use, and the code-switching differences between children and adults. A critical review of the peer-reviewed and scholarly literature to this end is followed by a summary of the research and salient findings in the conclusion.

Review and Discussion

Grammar of Code Switching.

The occurrence of code-switching -- the alternation of two languages both between and within sentences, has been demonstrated to be controlled by extra-linguistic (or social and situational) as well as intra-linguistic (or structural) factors. For instance, in her timely study on the issue, Poplack (2000) proposes two grammatical constraints which appear to operate on code-switching:

1. The free morpheme constraint which states that codes may be switched after any constituent in discourse, provided that constituent is not a bound morpheme; and,

2. The equivalent constraint, i.e. code-switching tends to occur at points in discourse where juxtaposition of two languages does not violate a surface syntactic rule of either language (Wei, 2000, p. 216).

According to Bakker (1997), there is apparently a clear typological dimension in code mixing wherein no or very little word-internal mixing that involves two fusional languages and massive word-internal mixing in pairs with one agglutinative language. Citing Myers-Scotton's (1993) proposed matrix language frame model to account for this type of code mixing, Bakker (1997) reports that: "Very briefly, this model states that there is one matrix language, in which foreign elements can be employed. That is, the matrix language provides the grammatical framework of the code-mixing utterance. The other language is called the embedded language. If the embedded elements are longer than one word, they should follow the grammatical system of the embedded language" (p. 180). Anything else should follow the grammatical system of the matrix language, including bound and free grammatical morphemes (termed "system morphemes" by Myers-Scotton, 1993). In this regard, Bakker (1997) suggests that, "This works neatly for many language pairs with different typological characteristics" (p. 180). Subsequent research has identified three types of code switching and code mixing: (a) alternation, (b) insertion, and (c) congruent lexicalization (which is intended to be associated with different social situations) (Bakker, 1997), as follows:

1. Alternation. This type of switching occurs in instances in which bilinguals start a sentence in one language and end in another. Both languages keep to their grammatical characteristics.

2. Insertional. In these cases, one language provides the grammatical framework, into which elements from another language are placed; the embedded elements are restricted primarily to stems or phrases. Bakker notes that, "This type can be associated with most of the language pairs that Myers-Scotton studied (1993)" (p. 180).

3. Congruent lexicalization. This is the most complex of the three types and shows an inextricable mixture of the two languages, whereby one cannot identify a matrix language and whereby the grammatical systems do not remain separate but have considerably converged. According to Bakker (1997), "This type may be rarer than the other two types" (p. 180).

It is these components, then, provide form the basis for the grammatical form of code switching: "In the composition of linguistic structures, content morphemes contribute whatever semantic and pragmatic messages the speaker wishes to convey and, in this initiating role, they project directions selecting the system morphemes which are the nuts and bolts -- the grammatical system -- which give the constituent its form" (Myers-Scotton, 1993, p. 255). There are some important informative clues that present themselves as well during the code-switching process. For example, "Code switching signals contextual information equivalent to what in monolingual settings is conveyed through prosody or other syntactic or lexical processes. It generates the presuppositions in terms of which the content of what is said is decoded" (Myers-Scotton, 1995, p. 54). By and large, code switching is an oral function and does not appear to be amenable to the written form: "Code switching is seen less in prose, and within that framework surfaces very little in scholarly writing. It is used most often in dialogue, many times in forms meant to be representative of linguistic variation or stream-of-consciousness style utterances" (Callahan, 2003, p. 12).

Constraints of Code Switching.

Notwithstanding the value of the framework for understanding what compels bilinguists to incorporate components of two or more languages during speech, there are some constraints that have been identified by researchers that should be taken into account as well. For example, three general approaches are apparent in past efforts to conceptualize constraints on code switching:

1. Researchers articulated constraints on switching from the surface (e.g. largely using linear ordering, form class, or size of switched material);

2. Some researchers proposed that constraints were driven by the same principles or rules formulated under current syntactic theories to explain syntactic structures within a single language;

3. Simultaneously and then more recently as well, others suggested that a major linguistic constraint on CS is related to clashes in subcategorization restrictions between the languages involved (Myers-Scotton, 1993).

A number of models have been proposed to resolve the grammatical constraints on intrasentential switching to determine whether the switch from one language to the other tends to follows or violate the grammatical rules of either language (Myers-Scotton, 1993). In this regard, the free morpheme and equivalence model Poplack (1980) and Sankoff & Poplack (1981) were the first to postulate linguistic constraints operating on intrasentential code-switching. They defined two such constraints:

1. The free morpheme constraint predicts that a switch may not occur between a bound morpheme and a lexical form unless the latter has been phonologically integrated into the language of the bound morpheme (i.e. borrowing); for example, the Spanish -- English code-switch flipeando is a well-formed Spanish form, whereas run-eando is not, because the phonology of run is unambiguously English and that of eando Spanish.

2. The equivalence constraint predicts that the order of sentence constituents immediately adjacent to and on both sides of the switch must be grammatical with respect to both languages simultaneously.

The equivalence constraint is illustrated in Figure 1 below, where the dotted lines indicate permissible switch points, and the arrows the surface relationships of the two languages; switches can take place at, but not between, the dotted lines:

Figure ____. Equivalence constraint rule in bilingual code-switching.

Source: Blanc & Hamers, 2000, p. 262.

The assumption behind the concept of equivalence is that it serves to facilitate the code-switching process. According to Blanc and Hamers, there can also be equivalence of categories (lexical elements, morphosyntactic features, and so forth) as well as the relations between categories (for example, word order); the very concept of equivalence is problematic, though. For example, there no precise match between categories in different languages (Blanc & Hammers, 2000). Furthermore, from a psycholinguistic point-of-view, a speech community may recognize categories from different languages as equivalent, whereas another community may not (Blanc & Hamers, 2000). In addition, the equivalence constraint assumes that the two switched languages share the same categories and makes no prediction about categorial mismatches: "If switches occur at sites where there is no structural equivalence between the two languages (when, for example, the two languages require a different word order), they sometimes involve omission or repetition of constituents" (Blanc & Hamers, 2000, p. 262).

Language Acquisition: Quantitative vs. Qualitative Differences.

According to Byram (2000), "Just like words, the voice (the way of pronouncing words and phrases, intonation and rhythm), gestures of various kinds, visual signs, posture, gaze, proxemic positioning and so on contain information which can be captured and used by the interlocutor in an appropriate and efficient manner, irrespective of the language and culture and in spite of certain quantitative and qualitative differences" (p. 446). In this regard, Meisel (2000) suggests that language acquisition studies have much to gain from investigating the simultaneous acquisition of two (or more) "first languages": "By analyzing the development of two linguistic competences in one individual, we may be capable of sorting out more easily to what extent the underlying logic of development is determined by the grammatical system to be acquired, or the particular way of human language processing as opposed to properties of the individual or of the communicative situation. There is, in fact, a steadily increasing amount of research in this area" (p. 344).

In his seminal work, Second-Language Acquisition in Childhood, McLaughlin (1985) reports that early research into language acquisition by preschool children suggested that interference between languages is not as inevitable or universal as was once believed. "Contrastive analysis, in its traditional form, was not able to account for the vast majority of errors that second-language learners made; in fact, learners from quite different language backgrounds appeared to make the same types of mistakes in the target language," he adds (McLaughlin, 1985, p. 14).

Since these early studies into language acquisition, other studies have shown that transfer from the first language does take place in the speech of children from certain first-language backgrounds and at certain times during the learning process. Therefore, McLaughlin cautions that, "It is an exaggeration to say that transfer from the first language is minimal and unimportant. The acquisition of phonological, syntactic, and morphological structures in a second language involves an interplay of both developmental and transfer factors. Transfer errors do occur and are extremely interesting for the researcher because of what they reveal about the learner's strategies" (p. 14). Nevertheless, the learner's first language tends to influence second-language acquisition in less direct and more restricted ways than it was once believed: "The evidence suggests that preschool children approach the task of second-language learning in much the same way they approached the task of learning their first language. Some authors speak of the reactivation of children's facility for language acquisition or of a creative construction process" (McLaughlin, 1985, p. 14).

A recent analysis of the linguistic adaptations that speakers make when transferring Spanish elements into the modern speech of inhabitants of Easter Island, Chile (known as Rapanui) by Makihara (2001) examined Spanish transfers and the mechanisms of adaptation at the levels of phonology, morphology, lexicon, syntax, and discourse. This study included a discussion of the phonological adaptation; application of Rapanui-bound morphemes; possessive class assignment; kin and emotion semantic fields; syntactic category crossing; the introduction of a modal construction of obligation, coordinating conjunctions, and an adverb of negation; and the use of Spanish elements as discourse markers and the indexicality they make possible. This author reports that, "Among lexical items, nouns are most transferable. Synchronically, the transferred Spanish elements are integrated through a combination of phonological, morphological, and syntactic means. Diachronically, they may go through a process of nativization requiring a certain amount of time to be accepted as part of the 'native' lexicon and grammar" (p. 191).

Children vs. Adult Code Switching.

All children are unique, of course, and the manner in which they go about acquiring first or second languages differs from individual to individual, just as with adults; however, there are some commonalties that exist that can provide some valuable insights into the general processes that are taking place. In this regard, McLaughlin (1985) advises that, "Children vary greatly in language learning -- whether it be a question of a first or a second language" (pp. 15-6). Early studies of language acquisition by children showed surprised researchers when they determined that children's language was both orderly and systematic. According to Bialystok and Hakuta (1994), "Some of this they discovered by analyzing the productions of children that, by adult standards, would be considered errors. But the errors clearly revealed that the children were actively formulating rules rather than behaving like parrots" (p. 29). In fact, as Bakker emphasizes, "There are no objective criteria to decide what is complex and what is simple. All natural languages are learned without effort by children" (1997, p. 24).

A large number of studies have investigated the order in which grammatical inflections were acquired by children, which was also investigated in first-language learners. By and large, this line of investigation showed that learners of a second language followed a fairly common sequence of acquisition (Bialystok & Hakuta, 1994). This finding held true for child and adult learners as well as for a variety of native languages and situations of language learning. To be sure, there were anomalies: for example, the English articles a and the continued to present problems for native speakers of many Asian languages (which do not have articles); likewise plurals and number agreement for speakers of Japanese, which does not have these grammatical rules; the findings, though, were consistent in showing that commonalities in second-language learning were attributable to factors that transcended native language characteristics: "Second-language learning was not a process of modifying what you already knew to arrive at the second language. Instead, it was quite simply 'language learning,' a process of constructing a new system from all our available human resources" (Bialystok & Hakuta, 1994, p. 31).

According to her recent analysis of discourse, interview, and observational data, Lam (2004) reports that a mixed-code variety of English is typically adopted and developed among young speakers and their peers around the globe to construct their relationships as bilingual speakers of English and Cantonese. "This language variety served to create a collective ethnic identity for these young people and allowed the girls to assume a new identity in speaking English that doesn't follow the social categories of English-speaking Americans vs. Cantonese-speaking Chinese in their local American context" (Lam, 2004, p. 44). According to Bialystok and Hakuta (1994), though, studies and empirical observations confirm that bilingual children and, of course, adults do not confuse their two languages. "Functionally, there seems to be some measure of differentiation early on, and this separation of languages by context, purpose, and interlocutor (conversational partner) is taken as evidence that the representation of the two languages in the mind must be distinct" (p. 115). Notwithstanding their level of proficiency, neither the first nor second language of bilingual speakers is precisely the same as either language for the respective monolingual groups (Bialystok & Hakuta, 1994).

These authors emphasize as well that there are differences in both vocabulary and syntax that can account for such differences because the bilingual speaker's two languages have been conceptualized in a single representation and therefore have had the opportunity to influence each other (Bialystok & Hakuta, 1994). According to these authors, "Bilingual speakers of French and English have different mental representations for both French and English from those of monolingual speakers of either language. In two languages represented by distinct systems, one would expect each language to stand as an independent structure, and the first language to be the same as its representation for a monolingual speaker of that language" (p. 114). Citing the findings by Green (1986) again, the authors also emphasize that transfer operates in both directions, and each language is influenced by the presence of the other (Bialystok & Hakuta, 1994).

Other evidence supports this position as well. For example, code-switching is a common feature of bilingual speech; however, how can children shift between languages that were stored separately so easily? Furthermore, there is a growing body of evidence that suggests bilingual children enjoy meralinguistic advantages over their monolingual peers. For example, according to the research by Bialystok and Hakuta (1994), "This means that bilingual children know more about language and have greater insight into its abstract structure. One could more easily explain a bilingual child's benefiting from enriched conceptions of language if the two languages were in close contact and influenced each other. Only a unified representation could achieve this" (p. 115).

The first possibility is illustrated in a representational model developed by Green (1986); this model is based on the notion that mental processes (e.g., the mental processes that find words to fit the speaker's ideas) can be both activated and inhibited. According to Bialystok and Hakuta, "The idea is simple but elegant. What is necessary is some kind of neurological switch, which Green calls the specifier, that has the responsibility of selecting which language is to be used. This system is so constructed that separate first- and second-language representations are attached to the same conceptual structure" (p. 116).

As a result, the conceptual system contains all of the meanings that the speaker knows and is the milieu in which active cognition, or conceptual processing, takes place. Producing or using language can occur in two separate ways:

1. In the simpler case, the linguistic representation initiates linguistic output directly, without going through an intermediate stage of conceptual processing. Words are produced directly from linguistic representations ("Alas, much linguistic output appears to be of this type!"); and,

2. In the more complex case, an idea passes from the specific language representation into the common conceptual system. From there, it must pass through the specifier to determine which language has the job of verbalizing the idea. The specifier sends out two messages -- an activation message to the language output center that has been selected for verbalization and an inhibition message to the language output center that is to remain mute (Bialystok & Hakuta, 1994, p. 116).