[Part III of the book, dealing with “Social evaluation” incorporates results from a series of field experiments that formed an essential part of the methodology. The experimental approach did not take hold in sociolinguistics in any way comparable to the studies of speech production that formed the basis of Part II. Though each of these experimental methods has had a history of replication and development, the studies that incorporate experiments are few by comparison with those that do not.

The simplest experimental approach involves the reading of texts and word lists, and this of course is not uncommon. Minimal pairs are one step more complicated, since they involve a comparison of production and perception – complicated by the need to label those perceptions. The self-report test (Chapter 12) is not difficult to prepare – a simple recording of the range of phonetic variants that have already been coded for the study of speech production. Linguistic insecurity tests (Chapter 12) need somewhat more preparation. The matched-guise experiments that are the subject of this chapter demand the most time and effort for the production of the stimuli. The resources and information needed to prepare matched guise recordings will usually require a good year of exploratory work. I hope that the results of this chapter will justify that effort, and reinforce the view that a successful study of the speech community must give considerable attention to normative behavior.]

We have completed our survey of social differentiation of five phonological variables in New York City, and we will now turn to the more obscure and difficult question of the subjective evaluation of the variables by our informants.