This project focuses on the issue of linguistic coherence, i.e. the question how much systematicity can be found in everyday linguistic heterogeneity. This is actually an old issue, however still causing much controversy today. The point of departure is that to address the issue thoroughly, the level of the individual has to be the prime focus of concern, contrary to the common practice of immediately aggregating over multiple speakers. The project will address the question to what degree individuals vary in coherence and how this variation can be explained. We will build a stylistically stratified corpus of spoken Netherlandic Dutch, which be analysed quantitatively adopting a bottom-up approach, charting all pronominal and inflectional variation (minimizing the risk of selection bias) and mapping covariance patterns in this variation. Coherence and interspeaker differences will be measured using entropy statistics, state of the art multivariate techniques and implicational scales.