The goal of this volume has been to offer an account of how knowledge of multiple languages is represented and accessed during intrasentential CS, the use of two or more languages in the same sentence. In what may be considered a companion volume ( Myers-Scotton, 1993), I discuss how the knowledge of multiple languages is exploited for socio-pragmatic purposes; that is, I consider the socio-psychological motivations for CS. Here I summarize the structural-constraints model, the MLF model, the main subject of this volume. I go on to suggest some ways in which the socio-psychological motivations and community norms for CS intersect with structural constraints.
Crucial generalizations about structural constraints on CS can be captured by recognizing the interplay of two hierarchies.
1. One of the languages involved in CS plays a dominant role. This language is labelled the Matrix Language (ML), and its grammar sets the morphosyntactic frame for two of the three types of constituent contained in sentences showing intrasentential CS, ML + EL constituents (those showing morphemes from the two or more participating languages), and ML islands (constituents composed entirely of ML morphemes). The third type of constituent, the EL island, is produced when ML morphosyntactic procedures are inhibited and EL procedures are activated.
2. The major organizing device which the ML uses in setting the frame is the division between system and content morphemes. The way in which content and system morphemes are differentiated in the MLF model becomes a theoretical claim about some of the crucial distinguishing features of lexical categories in all natural languages. Recall that properties regarding quantification and thematic-role status are used to make this distinction. The model claims that the feature [Quantification], viewed from the perspective of a system of logic, is a property of all system morphemes, since they pick out individuals across variables.