This paper discusses how collostructional analysis can be applied to the study of word-formation patterns. Drawing on a diachronic corpus study of German ung-nominalization and Infinitival Nominalization using the GerManC Corpus, it is shown that a cross-tabulation analysis comparing the frequencies of word-formation products to those of their respective bases can give valuable clues to the input language users rely on when abstracting a schema (i.e., a word-formation construction) from a quite heterogeneous array of instantiations.