Spanish/English codeswitching in fiction: A grammatical and discourse function analysis

The bulk of the vast body of literature on codeswitching treats oral codeswitching. The majority of studies on written Spanish/English codeswitching concentrate on poetry. Most of the studies that have been published on codeswitching in fiction deal with the sociolinguistic situations reflected by this practice; few attempt a structural analysis. This dissertation examines the general similarities and differences between oral and written codeswitching, and tests Myers-Scotton's Matrix Language Frame (MLF) model of structural constraints on a corpus of written data. The corpus consists of a total of thirty texts: novels and short stories containing Spanish/English codeswitching published in the United States between 1970--2000. The application of a model formulated for speech to a written corpus has implications for the relationship between spoken and written data and the use of the latter in linguistic analyses. Whether the patterns of codeswitching in the corpus support or fail to support the principles of the MLF model has significance in relation not only to what adjustments to this model may be necessary, but also to whether constraints formulated on the basis of oral codeswitching can predict written codeswitching. The analysis reveals relatively few counterexamples to the MLF model; the overwhelming majority of codeswitches found in these texts can be accounted for by Myers-Scotton's model of structural constraints. The most salient feature of the data is a high incidence of Matrix Language (ML) shifts at mid-sentence. Modifications are proposed to the original exposition of the model in regard to the flexibility of the ML and the constituent types possible.