This paper introduces the notion of meta model provenance. Meta model provenance helps to understand the origins of meta model elements, such as language concepts, attributes, or constraints. Thus, it should answer questions such as: where did this language concept come from? under which assumptions was it introduced? Among others, meta model provenance is intended to support the controlled evolution of languages and informed language (re-)design. In this paper, we focus on a goal-driven meta model provenance approach. This is one specific operationalization of the meta model provenance concept, which shows how goal models help to understand the origins of the elements of a conceptual modeling language. To illustrate our goal-driven provenance approach, we use a scenario from the electricity domain.