ON THE UNPREDICTABILITY OF CONTACT EFFECTS

SARAH GREY THOMASON
University of Pittsburgh (USA)

Historical linguists know that any search for deterministic predictions of language change is bound to fail, whether the focus is on internally-motivated change or on contact-induced change. But the urge to explain linguistic change is strong, and many linguists have proposed generalizations that make limited predictions about what can and can't happen in language history. Some of these generalizations, of course, are more successful than others; in particular, the ones that focus on tendencies rather than absolute constraints are more likely to be useful.

In language contact situations, the major predictors of possible linguistic results are social rather than linguistic. Linguistic factors are important but definitely secondary: there is solid evidence that they can easily be overridden by social factors that push in the opposite direction.

Specifying the crucial social factors, however, turns out to be difficult. Beyond the most obvious (and trivially true) social predictor -languages cannot undergo contact-induced change unless their speakers are, directly or indirectly, in contact with each other- only one social factor seems to yield a reliable constraint on the linguistic effects of contact: the presence or absence of full bilingualism among the speakers who introduce interference features into a language. Since you can't borrow what you don´t know, the range of interference features in any given case is constrained by speakers´ (lack of) knowledge of the source language. It is this fact that underlies the distinction between borrowing (into one's native language) and shift-induced interference (as a result of imperfect learning). This distinction in turn correlates with different linguistic results.

Other social factors are much harder to generalize about. In most cases, for example, a long period of contact is necessary to bring about structural borrowing, but there are exceptions; and the relevant sizes of the groups in contact, especially in cases of language shift, usually affect the linguistic outcome of the contact situation, but again there are exceptions. Such exceptions can be loosely ascribed to differences in speaker attitudes, but efforts to find reliable correlations across a wide range of contact situations between any specific attitudinal factors and specific linguistic results are probably doomed.

My goal in this paper is therefore not to offer predictive generalizations about ways in which attitudes determine the linguistic results of contact, but rather to show why it is so difficult to find any. First, I will contrast situations in which intense cultural pressure has led to profound changes in the pressured language -overall grammatical attrition (in language death) or gradual borrowing of most of the grammar- with situations in which the same kind and degree of pressure has left virtually no linguistic traces in the pressured language. Second, I will describe less extreme cases in which generally valid predictions about the kinds of effects to be expected in particular kinds of contact situations fail; one striking example here is the transfer of click sounds from Khoisan languages to Bantu languages in southern Africa, a change that was apparently unaccompanied by comparable levels of change in other grammatical subsystems.

Finally, I will emphasize the role of speakers' creativity in determining the linguistic outcome of contact situations (and even in monolingual contexts). Some linguistic effects can only be explained as the result of more or less deliberate decisions to combine two (or more) languages in new ways -most spectacularly in the emergence of such mixed languages as Michif and Mednyj Aleut, but also in cases involving less-than-full bilingualism, such as Halbdeutsch and pidgin/creole languages. Although motives for the development of such mixtures can sometimes be discerned, for instance a desire to develop a new language to go with a new ethnic identity, or the need to develop some means of communication, there seems to be no way to predict when speakers will decide to exercise their creative abilities in such a way, much less to predict what the specific linguistic outcome will be. Here too there are exceptions to generalizations that might at first seem obvious. So, for instance, new ethnic groups have certainly emerged without developing new mixed languages, and some pidgins have developed when the opportunity and motivation for full bilingualism / multilingualism might be supposed to have been present.

The paper will conclude with some comments on the status of explanations of language contact phenomena.