Theoretical framework

One of the characteristics of diglossia is the existence of linguistic convergence. In situations involving the functional separation of language varieties, speakers tend to modify their linguistic systems such that there is a level of one-to-one correspondence between elements of the coexisting systems (Gumperz and Wilson 1971: 154–166). Where some aspect of the linguistic system of one language variety is more complex than the other, there are consequences for trying to achieve this one-to-one correspondence.

Complexity may be defined in two ways. One system may make a greater number of distinctions than does another. Here, one consequence of convergence is that often the equivalent of a single form in a simple system may be two or more forms in a more complex system, one such form being common to both systems. Thus, in comparing the phonology of cognate lexical items of two language varieties, a form in the less complex system may be equivalent either to two or more cognate forms in a more complex one. The form in the more complex system will invariably predict its cognate in the less complex one, but not vice versa.

We make the following prediction about the relationship between phonological systems in the conditions of linguistic convergence which exist between JamC and JamE. If the simpler system, Variety A, has feature X and the more complex one, Variety B, has both the features X and Y, the initial hypothesis for speakers familiar with Variety A is that X in Variety A is equivalent to Y in Variety B. They thus convert all Xs to Ys in their effort to use Variety B. Later, with more exposure to B, speakers of A will learn that sometimes X in their native variety is equivalent to X in the target variety and only sometimes to Y. For speakers to know the difference requires lexical specification of individual items.

We make a second prediction about the relationship between the varieties in such conditions. Linguistic categories or variables often exist in pairs, e.g. the realization of segments equivalent to JamE /θ/ versus the realization of segments equivalent to JamE /ð/, or the realization of segments equivalent to JamC /ia/ versus those equivalent to JamC /ua/. Let us take the case of pairs of related linguistic variables, Variables 1 and 2, which, in Variety A have reflexes T and X respectively, and in Variety B, U and Y. We predict that in actual everyday usage of Variety A, only one of the two Variety A reflexes, e.g. T, will be consistently used. The other, X, will be used varying with Y, the variant associated with Variety B. Along similar lines, in the case of Variety B, only one of the two Variety B reflexes, this time Y, will be used consistently. The other, U, will vary with T, the form associated with Variety A. This is demonstrated in the table below.

The relationship between JamC and JamE presented in (1) represents a classic example of differential convergence.