A quantitative approach to the grammaticalization of discourse markers: Evidence from their sequencing behavior

This article takes a quantitative approach to the grammar of English two-part discourse marker sequences like oh well, you know I mean, etc. We investigate the internal ordering preferences of such sequences in spoken American English corpus data from the perspective of grammaticalization. From this perspective, the development of many discourse markers can be understood as involving a process of increasing syntactic de-categorialization (Hopper 1991) as the grammaticalizing element loses its original grammatical constraints and comes to function as a marker at the level of discourse. We test the hypothesis that discourse marker grammaticalization results in largely unconstrained ordering possibilities. Our analysis shows that, on the contrary, discourse marker sequencing is highly constrained. We interpret these constraints in terms of Auer’s (1996) model of discourse marker grammaticalization. Discourse marker sequencing is characterized by strong persistence of a marker’s original syntactic category and reflects its specific grammaticalization trajectory.