MATS THELANDER Code-Switching or Code-Mixing? 1. From a methodological point of view, the analysis of linguistic variation has to be adapted to the nature of this variation in each par- ticular case.' In a community where one is dealing with two or more substantially different languages in alternation it is possible, as a rule, to work on a MACROLINGUISTIC level. In such cases it makes sense to ask members of the community which language they would use in certain specified situations and, with a reasonable knowledge of the different languages, an observer has little difficulty in deciding in which speech variety a certain speaker is expressing himself at any given instant. According to this model, the speech repertoire falls into discrete (which does notexcludesome portion ofcommon features) and locally recognized varieties, codes, or systems. In communities that are traditionally monolingual, linguistic variation is also prevalent and a number of investigations over the last few years have demonstrated how this variation is socially structured and, from a functional point of view, analogous to a shift between markedly different languages. For settings of this later type, however, it has usually been felt more proper to apply a MICROLINGUISTIC approach. It is true that mention is sometimes made of sociolects, registers or styles, as if they were distinct systems of the repertoire, but at the same time the researcher often contents himself with pointing at a correlation of single linguistic variables with social strata or contexts. In many cases the awareness of this kind of variation is slight on the part of the speakers themselves. Due probably to an increase in geographical and social mobility in recent years and to an explosive development of mass communication. linguistic situations are now gradually emerging for which neither of these two models for description can be automatically accepted. Neutral- izing between existing varieties must be expected to take place, and the consciousness among the speakers of a discrete opposition between different varieties will become blurred as a cause or result of this process. A macrolinguistic view may then seem overpretentious whereas a micro-104 Mats Thelander linguistic approach tends to be somehow insufficient. This certainly is true of many settings where a local dialect has for some time been com- peting with the standard variety of the society at large (cf., for instance, Eriksson 1973). As has been argued by Blom and Gumperz (1972) conditions like these do not, on the other hand, necessarily bring about a complete state of flux. Although the linguistic distance between the two varieties in Hemnes, Norway, is shown to be relatively small, the authors conclude that for the time being there is little risk of a further merging of the two. Being associated with different value clusters, local vs. national, the codes themselves function as symbols in a sociosemantic system which plays a vital part in marking and maintaining the double identities of the Hemnes residents. As long as this distinction between identities is felt to be locally significant, the two codes are likely to remain segregated. 2. One of the aims of this paper is to make an attempt to apply Blom’s and Gumperz’ model of code-switching to a small Swedish community, which on the surface may show some resemblance to Hemnes. Being the most obvious linguistic consequence of a functional differentiation between varieties, a tangible code segregation will be regarded as the decisive indication of a possible Hemnes-type of bidialectalism. In order to acquire some objective measure of speech variation, the choice ofasuitable level of description is discussed in some detail. A compromise between a macrolinguistic and a microlinguistic approach will be proposed : an index of dialect level which incorporates a model of discrete speech varieties into a spectrum of continuous variation. 2.1. The site of investigation, Burtrisk, is a small community in a traditionally agricultural area of northern Sweden. It is situated 25 miles from the nearest town, Skellefted (see map, p. 15). Of the approx- imately 6,000 inhabitants of the district, the major part derive their livelihood from small-scale farming and forestry work. This group used to be much larger but because of the discontinuation and amalgamation of unprofitable small-holdings, a lot of people have been forced to abandon their native area. Strongly dominated by commercial and service trades, the community center has been affected by depopulation only indirectly. This village accommodates some 1,300 persons. A Burtrask resident is quite proud of his community, he has most of his friends and relatives close by, and he seldom finds reason to visit other places. Most things are at hand. Whether he lives in the villageCode-S witching or Code-Mixing? 105 or in the surrounding countryside, he calls and considers himself a burtraskare. One important symbol of this common identity AGAINST THE OUTSIDE WORLD is no doubt the dialect, burtrdskmdl, which differs considerably from the standard variety of Swedish. wITHIN THE COM- MUNITY however, burtrdskmal is not the usual designation of the local language. As in many other places in northern Sweden, people refer to the home dialect as bondska, “‘the language of farmers”. This is not altogether pejorative but it certainly suggests a dividing line in use and attitudes between the villagers and the rural population. Even disregard- ing the more formal situations, which for natural reasons are concentrated in the village, it is quite noticeable that the dialect has a comparatively weak footing among villagers. This makes it wise to modify the expecta- tions of an organized bidialectalism in Burtrask. Although, with few exceptions, people in Burtrask have a good understanding of both standard Swedish and the dialect, it is dubious whether standard-speaking villagers ever make personal use of the pure dialect or whether dialect- speaking non-villagers ever employ a mode of speaking which to any substantial extent differs from the dialect. To a direct question only 15 persons out of a sample of 52 declare that they do alternate between dialect and standard with the intention of keeping them separated. None of these are villagers. Another 23 persons admit that, depending on the circumstances, their speech may be more or less dialectal, but they prefer not to call it a regular shift. In this group quite a few villagers are represented. At any rate, the way of speaking certainly is not unimportant to the people in Burtrask. The following short extract from an authentic recording may demonstrate that. As a specimen of genuine dialect the text can also serve to illustrate the degree of difference between bur- traskmal (D) and spoken standard Swedish (S). The latter version is of course a construction: n ‘dei: vaf je ‘se: at om dem bera va “etm ‘dé:n en ‘de:l blir je ‘so at om dom bara er etm ‘den; some people get to be such that if they are outside the door only i *fjucton *de:ga se “cen: de int ‘at dem da dem i *fjucton “decgago ‘cen:ar man inta ifjen: dom do dom for fourteen days then you don’t recognize them when they xkomia ...va ‘jere sem ‘jo:r ‘he:? tkomier....va ‘eda som ‘joe: ‘de:? come back . .. how can this be? MEO Meo MED106 Mats Thelander 2.2 Empirical data for this study were drawn from linguistic analyses of 28 hours of recorded group discussions, supplemented by individual answers to a questionnaire from 52 out of those 56 informants who took part in the group discussions. With one or two exceptions the speakers were all born and residing in either the central village of Burtrask or in its immediate rural surroundings. The selection of informants can best be described as a ‘judgment sample” (Labov 1963: 284) within the age bracket 14 to 61 years. Informal group discussions were chosen for the recordings to secure a fair amount of intra-individual speech variation (cf. Gumperz 1972: 208). Each of the 14 group sessions lasted for two hours and included four speakers. To a large extent the groups were self-recruited (cf. Blom and Gumperz 1972: 426) and the discussions took place in a relaxed atmosphere in the home of one of the members of the group. All speakers knew that they were being recorded but to a greater or lesser degree they were ignorant of the exact motive for this.” For the purpose of generating a controlled shift of speech situation the informants, after one hour of talking on their own, were joined by a stranger introducing himself as member of the research staff. It seemed reasonable to expect that this introduction would make the speakers associate him or her with activities not belonging to the local sphere. To make sure that the informants did not refrain from using their dialect for reasons of intelligibility, care was taken always to engage a person from the north of Sweden to play the part of “stranger”. The topics under discussion varied from local gossip to world politics. 3. Linguistic variation on a microlevel is usually discrete in character. 3 The variable aj assumes (or is perceived to assume) either of the variants @q OF a,, normally nothing in between. To talk about discrete variation on a macrolevel as well (meaning that the speech alternates between the distinct varieties V) and V,) would presuppose a high measure of correspondence between different variables. One would expect variant Ap to be consistently co-occurring with variant bo, rather than with by at some instances and with an alternative form b, at others. This last- mentioned state of affairs, on the other hand, is something one has to take into account when examining the speech habits in a monolingual community. The question then is if ina community like Burtriisk one has to confine oneself to describing the variation as a continuum between the extremes of dialect and standard, or if, after all, the variation can be seen as a switch between a limited number of natural levels or discrete varieties. This latter solution of course represents a more structured view