Some recent approaches to language change within a constructional framework have been concerned with the development of procedural and of contentful constructions. This article considers how work in the tradition of grammaticalization has informed such approaches and points out some of the differences between the two research traditions. It highlights some of the problems around adopting key features of the grammaticalization tradition into a model of language structure which sees the language system as a network of conventionalized pairings of form and meaning, in which specific micro-constructions inherit properties from more general schemas. It also considers key issues in language change, such as directionality, the relationship between reanalysis and analogy, and the status of clines and parameters.