The article presents a problem of demarcation which hasn't been discussed in phraseology research so far: the differentiation between modified idioms and phrase schemata. With the help of a corpus based approach and on the basis of so called formulaic (ir-)regularities (i.e. phrasemes which deviate more or less from the synchronic free language use, e.g. auf gut Glück or seines (eigenen) Glückes Schmied sein), it will be shown that in certain cases the transition between modifications and phrase schemata is fluent. The method used in this article is a type of corpus analysis called slot analysis (based on the "Deutsches Referenzkorpus"/"German reference corpus") (Steyer 2013). On the one hand, corpus analysis allows for the illustration of such problems of demarcation. On the other hand, this method can be an important instrument in the operationalization of an exact differentiation between modified and schematic phrasemes on the basis of quantitative analysis of real language use.