ith a large number of quite diverse Dutch verbs, the language user has a choice to express the object argument as a direct object (1) or a prepositional object (2). (1) Dirk zoekt zijn sleutels. (2) Dirk zoekt naar zijn sleutels. 'Dirk is searching his keys.' (1) De koning regeert zijn volk. (2) De koning regeert over zijn volk 'The king rules his people.' This project will investigate what makes the language user opt for (1) or the alternative in (2). In doing so, we will consider a wide range of possible factors: e.g. a subtle difference in meaning, a number of linguistic parameters or even language-external factors, such as the nationality of the language user and formality of the situation. To find these factors, we will look for patterns in huge digital collections of natural Dutch language material. This will involve the use of statistical models such as logistic regression, as well as distributional techniques, which infer an expression's meaning from its context. Although this project's main aim is to advance (usage-based) theoretical models of how language functions, its findings will also serve a direct use. Until now, second language learners of Dutch can only be told that the choice between (1) and (2) is a matter of 'feeling for the language'. This project however, will quantitatively show what determines this linguistic feeling.