People who migrate to another country and start to use the new country’s majority language, frequently experience changes in the way they use their native language, oftentimes due to transfer from their second language. This transfer tends to occur to varying extents for different construction types, with transfer generally occurring more frequently in the case of lexically specific constructions (e.g., words or word combinations) than in the case of schematic constructions (e.g., word order patterns). To test this experimentally, this project compares transfer from Dutch to German by native German speakers living in the Netherlands for lexically specific constructions (LVCs, for example, *Wecker setzen instead of Wecker stellen transferred from the Dutch wekker zetten, to set an alarm) and schematic constructions (‘om’-constructions, for example, versuchen *um instead of versuchen – transferred from the Dutch proberen om, to try to). The results suggest that there are differences in the transfer patterns of the two construction types. In the case of the LVCs, speakers seem to transfer each construction separately, for example, whether or not they use the transferred *Wecker setzen is not related to their use of other transferred LVCs. In the case of schematic ‘om’-constructions, on the other hand, transfer is related across items, with some speakers never using any transferred ‘om’-construction and others using these constructions in the majority or even all cases. This suggests that speakers have formed a schematic representation of the transferred pattern that they can use to either (a) apply it across all or at least multiple instances of the construction or (b), when aware that the resulting constructions are ungrammatical in German, avoid its use in all cases. As such, our results show that a higher level of schematicity can lead to both less and more language transfer.