Interactional otherness: towards a redefinition of codeswitching *. (Main Articles).

Abstract

The concept of codeswitching has traditionally been understood, from an external observer's perspective, to mean any occurrence of two languages within the same conversation. Our position is that language alternation should be observed from participants' own perspective. From this perspective, many distinct types of language alternation can be identified, namely "language alternation itself as the medium" "medium repair" (Gafaranga, 1999, 2000) and functional deviance from the medium. In turn, in this last possibility, two situations can be found: "medium-switching" and "medium suspension." However, for practical reasons of research continuity, we suggest to keep the term "codeswitching" to refer to both cases for both are instances of interactional otherness. Thus, our "respecification" is that codeswitching is, not any occurrence of two languages within the same conversation, but rather any instance of deviance from current medium which is not oriented to (by participants themselves) as requiring any repair.

Key words

codeswitching

medium-switching

medium-suspension

1 Introduction

When Gumperz initiated the study of the phenomenon he referred to as codeswitching (1), he defined it as "the juxtaposition within the same speech exchange of passages of speech belonging to two grammatical systems or subsystems" (1982, p. 59, our emphasis). Today, researchers' opinions about the adequacy of this definition are divided. While some (e.g., Myers-Scotton, 1993) use the term "codeswitching" as if it were synonymous to "language switching," others (e.g., Auer, 1984a, 1988, 1995) call for a distinction among the different language contact phenomena. In this respect, within the broad category of language contact phenomena, Auer distinguishes code-alternation from what he refers to as the new code. In turn, within code alternation, he distinguishes codeswitching from transfer. However, even those who use the term "codeswitching" to refer to any instance of language-alternation feel the need to distinguish among its different types. For example, Gumperz distinguishes situational codeswitching from metaphorical or conversational codeswitching. Similarly, Myers-Scotton distinguishes marked and unmarked codeswitching. Briefly, the need for a clear definition of what counts as codeswitching remains a concern for researchers. The aim of this paper is to contribute to this debate.

At first, the debate as to what counts as codeswitching may seem to be merely terminological. However, if the above categories are looked at closely, the debate turns out to be one of substance. The above categories are based on different theoretical perspectives on social interaction and on different views about the notion of language and that of code in social interaction. Further, they reflect different epistemological orientations. Gumperz and Myers-Scotton, for example, adopt an "identity-related explanation" while Auer adopts a sequential perspective (Sebba & Wootton, 1998). Gumperz and Myers-Scotton are interested in the social values of languages and "the social motivations for codeswitching" (Myers-Scotton, 1993) while Auer proposes a "conversation analytic approach" (1988). On the level of the views about language and code, they all use the notion of code, but they seem to understand it differently. For example, in Myers-Scotton, the notion of code and that of language seem to be synonymous. Auer, on the other hand, concedes that language and code need not be the same. According to Auer, language alternation itself can be a "(new) code" (1984a, p. 9). At the epistemological level, Auer argues for an approach where claims are based on participants' own perspective as revealed in the local organization of talk. By contrast, Gumperz and Myers-Scotton put forth a more ethnographic approach, whereby analysts' knowledge of the social values of languages in the community plays a crucial role in the interpretation of the data. This paper adopts the perspective initiated by Auer whereby social interaction is looked at from an organizational perspective and the notion of code and that of language are seen as different.

Although, in principle, Auer seems to make a distinction between language and code, in practice, he reflects the monolingual bias that has so far characterized studies of language alternation. This bias can be seen in the overarching principle of preference for same language talk that underpins Auer's analysis. According to Auer,

 
   (...) in many bilingual communities, there is preference for same language 
   talk, codeswitching (...) runs counter to this preference--which of course 
   only heightens its signaling value--while transfer is neutral vis-a-vis 
   questions of negotiating the language-of-interaction. (Auer, 1984a, pp. 
   28-29) 

Gafaranga (1998, 1999, 2000) challenges this monolingual bias and proposes a "respecification" of the above preference as preference for same medium talk. As Gafaranga argues, the code in bilingual interaction need not be monolingual. It can also be bilingual (also see Gafaranga and Torras, 1998, 2001). Therefore, our aim in this paper is to revisit the concept of codeswitching starting from the perspective whereby the notion of language and that of code are viewed as different.

The discussion in this paper will be conducted on the basis of three sets of data, namely Kinyarwanda-French data, Catalan-Castilian data and Catalan/Castlian-English data. Space does not allow us to describe these data sets in any detail (see Gafaranga, 1998 and forthcoming, Torrass, 1998 and 1999 respectively). However, some important differences between the sets need to be pointed out. As Myers-Scotton (1993) observes, different patterns of language alternation can be observed in different communities. More specifically, according to Myers-Scotton, unmarked codeswitching is unlikely in sociolinguistic contexts characterized by more or less open linguistic conflicts. This seems to account for the differences we observe between our Kinyaranda-French data and Catalan-Castilian data sets. The Kinyarwanda-French data come from an African postcolonial context (Rwandese societal bilingualism) while the Catalan-Castilian data come from Catalonia, a context which has widely been described as characterized by the conflict between Catalan and Castlian (see Woolard, 1989, for example). As a consequence, in the first set of data, frequent alternation between the two languages is observed and it obtains both within turns and between turns. A main issue in this type of data is whether one can speak of one language constituting the "base language" of a particular conversation. On the other hand, in the second set of data, collected in service encounters in the Barcelona area, language alternation is less frequent and is mainly of the interturn type. Secondly, in the literature, language alternation has been investigated in contexts where only two languages, generally community languages, are available to participants. Our third set of data, also collected in the Barcelona area, involves three languages (Catalan, Castilian and English), one of which is used as an international language. Here too, language alternation is mainly of the interturn type. Finally, language alternation has been investigated in informal conversations among bilingual speakers. This is the case in our first set of data. Participants, all Rwandese refugees in Belgium, are relatively long-time friends. Our second and third sets of data, on the other hand, consist of service encounters, of institutional talk in "non-formal settings" (Drew & Heritage, 1992, p. 27), and, in most cases, these are first time encounters. Using these three sets of data, we hope to provide a broader picture of bilingual conduct than would have been possible if only one set of data had been used.

2 Some note-worthy situations in talk among bilingual speakers

In order to define codeswitching as a specific type of language alternation, it is necessary to place it within a global view of the phenomena that may occur in bilingual interaction. In this section, we will illustrate some of the phenomena we have observed in our data sets. First, among bilinguals, talk may be conducted in one language only. In most work on language use among bilingual speakers, this situation is usually not taken notice of as deserving any account. Rather, it is taken to be the norm, the normal way of talking among both monolinguals and bilinguals. This is what we have referred to above as the monolingual bias. In this paper, we will not address this phenomenon per se for the practical reason that our focus is on codeswitching. Elsewhere, we have accounted for this situation in terms of the monolingual medium (Gafaranga & Torras, 1998, 2001).

The second situation refers to cases where participants use both of their languages "without any obvious motivation" (Labov, 1972). Unlike monolingual language use, this way of speaking is often noticed, not only by researchers but also by community members themselves. Among researchers, this practice is commonly referred to as "codemixing" and more technically as "unmarked codeswitching" (Myers-Scotton, 1993) and as a "new code" (Auer, 1984a, 1998). Community members themselves often assign labels to this type of talk and, by doing so, confirm not only their noticing but also the nature of this way of speaking as a "stable" (Garfinkel, 1963) practice. Examples of such labels are Franglais, Finyafrancais, Spanglish, Yanito, and so forth. Here again, the reason why this type of language use is noticed is precisely the monolingual bias we have mentioned above. This situation is taken note of because it deviates from the assumption that the norm is to use one language only.

Although researchers take note of this type of talk, because of the very same monolingual bias which occasions their noticing, they often fail to raise, let alone address, a crucial issue regarding this mode of speaking. Researchers do not raise the issue of the very possibility of this type of bilingual talk, of its …