This chapter focuses on a traditional issue of word formation, namely synthetic compounds. The three basic approaches to the question will be reviewed and analyzed with the help of a large text corpus. The latter is of paramount importance for the analysis because dictionaries usually are not reliable for investigating highly productive word formation patterns and especially German compounding. It is shown that purely syntactic approaches do not cope well with the data, whereas a lexical approach like Construction Morphology is able to grasp the fine-grained distributional properties displayed by compounds. Furthermore, the corpus-based analysis allows us to shed some light on the complex network of semantic properties guiding the selective solidarity between deverbal head and nominal modifier by representing argument structure as a bundle of Dowty’s base-roles.