A central concern in linguistics is assessing the linguistic competence of individuals or groups. Formal linguists usually accomplish this by the study of intuitions with little regard for observed usage, while survey sociolinguists usually depend on observations – especially the data of “spontaneous” interviews – with little regard for intuitions. In this paper I argue that survey sociolinguists need to make greater use of repeated recordings and elicited intuitions.

The existence of this need is illustrated in most detail by an attempt to replicate an earlier implicational analysis of pronominal variation in the Guyanese creole continuum. It is shown that with repeated sampling and the inclusion of elicited intuitions, the discontinuities on which implicational scaling depends disappear almost entirely. With a clearer idea of what speakers can say, however, the sociolinguistic interpretation of what they do say in the spontaneous interviews and recordings is rendered more reliable and revealing.

In the conclusion, some of the theoretical implications and methodological difficulties involved in extending the use of repeated recordings and elicited intuitions in sociolinguistic surveys more generally are discussed. (Sociolinguistic survey methodology, variation, style, implicational scaling, creoles)

Quantitative analyses of large data sets make use of both linguistic and sociological categories in sociolinguistic studies. While the linguistic categories are generally well-defined and there are sufficient tokens for further definition based on mathematical manipulation, the social characteristics such as socioeconomic class or ethnicity are neither. The familiar problem of grouping speakers by such sociological characteristics prior to quantitative analysis is addressed and an alternative solution – principal components analysis – is suggested. Principal components analysis is used here as a heuristic for grouping speakers solely on the basis of linguistic behaviour; the groups thus defined can then be described according to sociological characteristics. In addition, by naming the principal components, the major linguistic and social dimensions of the variation in the data can be identified. Principal components analysis was applied to vowel variation data collected as part of a sociolinguistic survey of English in Sydney, New South Wales, Australia. (Sociolinguistics, variation studies, quantitative methods in linguistics, dialectology, Australian English, role of migrants in language change)

In 1977, a bilingual education project began in rural areas of Puno, Peru, as a direct result of Peru's 1972 Education Reform. This paper presents results of an ethnographic and sociolinguistic study comparing Quechua language use and maintenance between: 1) a bilingual education school and community, and 2) a nonbilingual education school and community. Classroom observation indicated a significant change in teacher–pupil language use and an improvement in pupil participation in the bilingual education school. Community observation and interviews indicated that community members both valued and used their language. Yet the project has had difficulties expanding or even maintaining its implementation. (Quechua; Puno, Peru; Peru; Andes; bilingual education; classroom language use; ethnography; sociolinguistics; community development; language planning; language maintenance; educational policy)

In applied sociolinguistics in West Germany the notion has been influential that the indigenous working class is separated from the middle class by a linguistic barrier and thus is at a linguistic disadvantage, as well as suffering other forms of disadvantage. The paper places this view in its context within German work on language and society, examines it critically, and outlines why, in the author's view, it is of questionable validity. (Sociolinguistics, dialectology, education, German)