The recent revelation that there are correlated period derivative and pulse shape changes in pulsars has dramatically changed our understanding of timing noise as well as the relationship between the radio emission and the properties of the magnetosphere as a whole. Using Gaussian processes we are able to model timing and emission variability using a regression technique that imposes no functional form on the data. We revisit the pulsars first studied by Lyne et al. (2010). We not only confirm the emission and rotational transitions revealed therein, but reveal further transitions and periodicities in 8 years of extended monitoring. We also show that in many of these objects the pulse profile transitions between two well-defined shapes, coincident with changes to the period derivative. With a view to the SKA and other telescopes capable of higher cadence we also study the detection limitations of period derivative changes.