ABSTRACT
Competition between relativisers has persisted over time and across varieties of English. We compare the distribution of and constraints on relativisers in speakers from two age groups and three areas of Auckland, New Zealand: Titirangi (stable and predominantly white/Pakeha), South Auckland (ethnically mixed for some time), and Mount Roskill (formerly predominantly white/Pakeha but now ethnically mixed). Results provide little or no evidence that relativiser selection is socially constrained variable; instead, it is linguistically constrained, as Cheshire suggested is true for most syntactic variables. Our analysis of a hitherto unattested effect of the semantics of the antecedent provides a principled basis for the use of each relativiser.
