Learning argument structure constructions is believed to depend on input properties. In particular, in a cued production task, verb production within each construction has been shown to depend on three input factors: frequency of a verb in a construction, contingency of verb–construction mapping, and verb semantic prototypicality. Earlier studies have estimated these values from a language corpus, without accounting for variation in the input to individual learners. We use a computational model to control for such variation, and our results replicate those reported for human learners. The second issue that we address relates to different ways of representing constructions: while the earlier studies employ form-only representations, we run an additional analysis for form–meaning representations. Again, the results show the impact of all three input properties on the verb production, but their relative impact depends on the representations used.