A study investigated the general occurrence of English lexical elements in the recorded speech of 285 Dutch pre-World War I immigrants and their descendants in Massachusetts, Iowa, and Wisconsin. The problems encountered in analyzing a massive data corpus and the methods used to resolve these problems are the focus of the paper. The difficulties of defining code-switching versus borrowing as contrasted with other language contact phenomena are first discussed; next the process of narrowing and operationalizing the characateristics of code-switching is described; and finally it is shown how these characteristics will be applied in computerized form to the data. Results from the "code-switching marker" analysis, used for both single and multi-word items, suggest that the phenomena attending occurrence of English elements can be an important source of information in code-switching analysis, particularly for multi-word elements, and that the categories of analysis must be refined further to determine loan items. Criteria of frequency of occurrence across speakers must be an important additional source of information in code-switching analysis because loans are by definition part of the vocabulary of the speech community as a whole and must occur in the speech of more than a single speaker. Overall, investigation of code-switching and borrowing is seen as complex. (MSE)