Sociolinguists (most recently Labov 1982 and Trudgill 1986) have assumed that dialect is stabilised by adolescence. Recently Montreal French (MF) data has been collected by a group of researchers headed by Pierrette Thibault and David Sankoff which permits the comparison of the speech of individual speakers in 1971 and 1984. The comparison of these data seriously challenges Labov's and Trudgill's theories. Perhaps Giles' Speech Accommodation Theory (SAT), and recent work by Thibault (1983), help explain how such disproof of the generally accepted theories an arise.

This paper initially summarises five questions which now figure quite prominently in quantitative sociolinguistic discussions of speaker classification: (1) how far are speaker categories emic or etic? (2) do they encode local inter-speaker relationships? (3) can they combine with accounts of interactional language use and learning? (4) do they handle flexibility and multiplicity as features of group membership? and (5) are interactive and referential group affiliation analytically distinguished? It considers the ways in which quantitative research has handled some of these but not others, and suggests that (4) and (5) currently present the greatest empirical challenge.

A methodology capable of overcoming (4) and (5) is then outlined and illustrated with a case study which investigated the distribution of two phonological variables across one multi-ethnic adolescent peer group. Its relation to (1), (2) and (3) is closely examined, and it is suggested that in comparison with other empirical endeavours, this approach does significantly extend the methodological repertoire of quantitative sociolinguistics.

This paper reports the results and conclusions of an auditory study of local Cambridge English, focussing on the effect of style on the vocalisation of /l/. This sociolinguistic feature has been described as a feature marking south eastern varieties of British English, and as a connected speech process in its sensitivity to variation in speaking rate. The study concludes that this feature is socially salient in local Cambridge English.