Starting from a data-driven approach, the current paper compares the BNC1994 spoken to the BNC2014. We first narrow down possible research questions due to differences in the compilation and transcription of the two BNC generations. Then we investigate three robustly detectable changes at the level of lexis and morphosyntax: (1) gender and class differences, (2) the increase of be- and get-passive constructions and -ing forms from the verbal domain, and (3) the in- crease of noun compounds from the nominal domain. We also focus on the so- cial context in which linguistic changes are embedded: which noun compounds particularly increase; which words are overused by which gender or social class? Technology seems to be a driver in the further advance of the construction of noun compounds, and strong swearing seems to have decreased between 1994 and 2014.