Language mixing and language fusion: when bilingual talk becomes monolingual

Language mixing and language fusion: when bilingual talk becomes monolingual

One. Introduction

After a time of intense and partly controversial discussion, the topic of "mixed languages" seems to have settled on a definition of mixed languages as languages with a "split ancestry". There is no agreement, however, on whether these mixed languages should be seen as strictly delimited from "unmixed" languages (which just show some 'ordinary' borrowing) or whether they are merely extreme cases on a continuum whose other extreme is 'ordinary' borrowing as it occurs in every language. The first case, advocated most strongly by Bakker, stipulates that mixed languages "must have numerically (roughly) equal and identifiable components from two (or more?) other languages"; only in this case would the above definition hold true in a strict sense. If the number of components from the two languages are not "(roughly) equal", a mixed variety may still be said to have language A as its (main) ancestor, even though it contains any number of elements from another language B. The continuum view, on the other hand, argues that there is no clear boundary between mixed languages and 'ordinary' borrowing. Mixed languages simply represent a rare case of massive borrowing from one language to another. This view implies that the split ancestry definition must be gradient: A mixed variety may be related to two ancestor languages, but in most cases, one of these ancestor languages will be the dominant (real?) one.

I will follow the gradient view of mixed languages in this paper and argue that there is nothing special or extraordinary about mixed languages from a grammatical point of view. The individual linguistic processes involved are those we know from language contact in general. What is special in the radical cases of mixed languages is their sociolinguistic status and history, i.e. the circumstances which led to such radical borrowing, sometimes within a

Language mixing and language fusion

short time-span. Very often, this happens in situations of language shift and language obsolescence, in which elements of the former group language are maintained by fusing them with the dominant, out-group language. As a rule (but not always) the sociolinguistically dominant language is also the structurally dominant one, i.e. the new variety is genetically speaking a variant of the shifted-to language, rather than a descendent of the lost or endangered language.

Looking at mixed languages as an extreme case of borrowing implies a shift from the result (a mixed language) to the process of their emergence (extensive borrowing). In order to underline this shift of perspective, suggest the term fusion for the processes of extensive borrowing of elements from another language into the receiving language, and the term fused lect to refer to its outcome. While the usual suspects for radically mixed languages are a few, partly insufficiently documented varieties whose status is often disputed due to a simple lack of data, the focus on fusion as a process opens up the possibility of investigating examples of partial fusion as well. In most of these cases, the ancestor language question will not be central; they can clearly be attributed to one ancestor language much more strongly than another one. What is central is the structural conventionalization and indeed grammaticization of bilingual mixing patterns.

The main argument of this paper is that the basis of language fusion is language mixing (often also called code switching), a phenomenon which we need to separate strictly from "mixed languages" (in the above sense of fused lects in their extreme form), and which has been abundantly documented in bilingual studies. Language mixing is a surface phenomenon which consists of the juxtaposition of two languages. It will be argued that most cases of fusion originate from such mixing (even though other routes cannot be excluded in principle). This leads to the most important question: What kind of surface patterns of bilingual talk become part of the language structure of a fused lect?

The main argument I will put forward here is that language fusion is the result of the conventionalization of the two basic strategies of insertional mixing. These two basic mixing strategies have been amply documented in studies on bilingual talk, where they may occur in isolation or in combination. The first strategy, called minimal insertion here, consists of inserting stems (or uninflected words) from one language into the grammatical frame "matrix" of the other language without any accompanying grammar. In this case the lexical material from the inserted language is accommodated completely to the grammatical structure of the matrix language. (Phonology and phonetics are not necessarily affected.) The second strategy, called maximal insertion or "embedded island" insertion here, prefers to bring the other-language item(s) into the matrix frame together with the accompanying grammatical markers (affixes or grammatical words) and often extends to the phrase level.

I will argue that these two strategies are the equivalents of the two main patterns of language fusion which can account for the structure of radically fused ("mixed") languages. Minimal insertion in mixing is the basis of the structure of "symbiotic" fused lects, in which the fusion affects the lexicon of one language and the grammar of another. The grammar is entirely in language B, and lexical items (usually stems) are selectively inserted into this pattern (examples are Media Lengua, or Para-Romani). Maximal insertion, on the other hand, accounts for the second group of fused lects in which two grammars merge,

The idea that fusion arises out of mixing is not new. It has sometimes been criticized on the basis that "no documentation of a transitory phase between the supposed code switching behavior preceding the mixed languages" existed, as Bakker, one of its strongest opponents, writes. In the meantime, however, there is historical evidence for such a transition. In this paper, additional evidence will be given by approaching the issue in a somewhat indirect way by showing that the structural similarities between language mixing and fused varieties are so substantial that it is highly likely that the former provide the basis of the latter. Intermediate stages of fusion will be presented which provide a better understanding of how fused lects emerge.

With regard to the overarching topic of this volume, it should be noted that the notion of split ancestry and hence the concept of language fusion is theoretically speaking orthogonal to that of family resemblance, i.e.: language fusion should occur regardless of whether the two languages in contact are genetically related or come from different language families. However, a look into existing research on "mixed languages" shows that by far the majority of cases discussed so far come from language contact between unrelated or genetically hugely distant languages. This raises the question of whether the same concept of fusion can be applied to closely related varieties as well or whether fusion contradicts family resemblance. I will briefly (and somewhat speculatively) address this question in the last section of this paper.