siSwati-English code-switching: the matrix language principle and linguistic constraints

Abstract

Sankoff & Poplack and others claim that code-switching (CS) is governed by two universal constraints, the Free Morpheme Constraint and the Equivalence Constraint. These claims have been questioned in Choi, who claims that CS is governed by the general rule switch alpha: the underlying property of the specific language-pair constraints is particular language-pair integrity, in which the characteristics of a certain language-pair are not violated. Drawing on siSwati-English CS, I argue in this article that Choi's claim is a replica rather than an alternative to Poplack's proposals and should be abandoned. The siSwati-English facts rather provide support for the idea of a matrix language principle proposed in Kamwangamalu and Myers-Scotton & Azuma. In essence, the matrix language principle says that in a CS structure any element can be code-switched provided the morpho-syntactic integrity of the matrix language, and not of the language-pair involved, is preserved.