peer reviewed ; This article illustrates some of the opportunities and challenges of pursuing a big data approach in linguistic research. To do so, we investigate the diffusion of the loan verb pimpen ‘to fancify’ in Dutch based on Twitter data. First, we focus on the derivations of the verb (e.g.: terugpimpen ‘to pimp back’, herpimpen ‘to repimp’, etc.) and plot the diversity of these forms through time, using the Chao-Wang-Jost estimation of Shannon entropy. We follow this up with an alternation study that compares pimpen not only to its ‘native’ alternative opleuken, but also its most frequent derivation oppimpen, using multinomial regression. It is found that, while pimpen’s early expansion in Dutch has proceeded at breakneck speed, resulting e.g. in a plethora of derivations that has so far gone undetected, its current momentum seems to be waning.