Error, feature, (incipient) change – or something else altogether? On the role of low-frequency deviant patterns for the description of Englishes

All corpus linguists routinely come across language use that may, at first sight, appear to be erroneous: planning errors, slips of the tongue/pen or, in the case of contact varieties, evidence of interference from a substrate language. These apparent ‘errors’ are of potential interest because they may, in fact, be on their way of becoming a ‘feature’ of a contact variety or be instances of ongoing change. With respect to change in previous stages of English, we can decide in hindsight which of the three possible scenarios we are dealing with. When it comes to variation in present-day English and possible (ongoing) change, matters are more complicated. In this paper, I present a case study on an unusual auxiliary-participle combination (be been), which has the potential of being an (emergent) feature of contact varieties or an instance of (incipient) language change. I compare two kinds of evidence: acceptability judgements and metalinguistic comments, on the one hand, and data collected from a broad range of corpora, on the other hand. The findings are used not only to discuss grammaticality and norms, but also to reflect on the methodologies employed in the description of World Englishes today.