Introduction

Dialects of the same language are known to vary systematically with respect to the proportional frequency with which different syntactic constructions are used productively. However, many syntactic variables are relatively rare in spoken interactions of the kind elicited by sociolinguistic interviews, particularly when the variants in question are stigmatised within the community in which they are present but also because of the open-endedness of the syntactic component. Relic features like ‘for-to’ complementisers, for example, are not only restricted to older generations of speakers, but their frequency in interviews even within this social group is also delimited by the fact that complementiser constructions are only one amongst numerous structural possibilities for conveying grammatical and pragmatic meaning. Due to the low token frequency of such variants, the investigation of large-scale dialectal variation within the syntactic component has increasingly come to rely on the collection and analysis of introspective judgements. This reliance has initiated an important discussion about the linguistic status and empirical appropriateness of judgement data within the field of dialectology. A number of scholars (Schütze 1996; Cowart 1997; Cornips and Poletto 2005; 2008) have drawn our attention to the fact that, if proper care is taken to control for potentially interfering, though independent, linguistic constraints – for example, lexical frequency/familiarity, pragmatic plausibility and sentence length – as well as for extra-grammatical factors – such as the social profile of the speakers, fatigue, memory limitation and ordering effects – native speakers can indeed be found to produce systematic patterns of acceptability ratings when using these methods.