Converging variations and the emergence of horizontal links

The guiding question of this paper is how (horizontal) connections are established when new items enter the network of constructions. It presents a quantitative, corpus-based study of the development of to-contraction (e.g. want to > wanna) in American English since the 19th century. From a plethora of earlier forms, gonna, wanna and gotta emerge, first as representations of phonetic reduction, but in time claiming their place as newly emerged, separate nodes in the constructional network. As their frequency increases, their usage patterns (relative to the full form) become increasingly similar. I propose that this marks the emergence of a horizontal link, which can be described as an emerging “metaconstruction”. The study discusses the status of these forms as either phonetic variants, potential allostructions, or independent form–meaning pairings, and provides first-hand evidence for emergent connections via mechanisms of analogy in language. Moreover, it makes an argument that (changing) usage patterns reflect (changing) constructional links and degrees of entrenchment.