ABSTRACT
A brief example will illustrate the method. Perhaps the bestknown exponent of the quantitative paradigm, William Labov, carried out a famous study in three New York City department stores (Labov 1972a). He was interested in the linguistic variable (r) - that is, the fact that in New York City r in postvocalic position can either be pronounced [ l] or not at all (these alternatives are known as variants). Labov asked sales staff in the three stores questions which elicited the answer fourth floor - a phrase in which the variable appears twice. Every time a subject pronounced the [ l] variant he gave them a score of 1; each time the zero variant was produced he awarded a score of o. He was thus able to arrive at an overall numerical score for the speakers in each store. The stores he had chosen were ranked by social status from a high-class establishment to a fairly downmarket one. This ranking was reflected in the scores for (r): in New York City it is prestigious to use the pronunciation [1], and Labov found that the higher the status of the store, the higher the score of its staff for (r). A more detailed study of New York City speakers confirmed that the variable (r) was 'socially stratified', that is, a speaker's (r) score could be correlated with their class.
