Aims. Every 5.5 years, η Car's light curve and spectrum change remarkably across all observed wavelength bands. These so-called spectroscopic events are most likely caused by the close approach of a companion. We compare the recent spectroscopic event in mid-2014 to the events in 2003 and 2009 and investigate long-term trends. Methods. Eta Car was observed with HST STIS, VLT UVES, and CTIO 1.5 m CHIRON for a period of more than two years in 2012−2015. Archival observations with these instruments cover three orbital cycles and the events of 2003.5, 2009.1, and 2014.6. The STIS spectra provide high spatial resolution and include epochs during the 2014 event when observations from most ground-based observatories were not feasible. The strategy for UVES observations allows for a multidimensional analysis, because each location in the reflection nebula is correlated with a different stellar latitude. Results. Important spectroscopic diagnostics during η Car's events show significant changes in 2014 compared to previous events.While the timing of the first He ii λ4686 flash was remarkably similar to previous events, the He ii equivalent widths were slightly larger, and the line flux increased by a factor of ∼7 Conclusions. The basic character of η Car's spectroscopic events has changed in the past two to three cycles. The ionizing UV radiation dramatically weakened during each pre-2014 event but not in 2014. The strengthening of He i and N ii emission and the weakening of the lower-excitation Hα and Fe ii wind features in our direct line of sight implies a substantial change in the physical parameters of the emitting regions. The polar spectrum at FOS4 shows fewer changes in the broad wind emission lines, which may be explained by the latitude-dependent wind structure of η Car. The quick and strong recovery of the He ii emission in 2014 supports a scenario, in which the wind-wind shock may not have completely collapsed as was proposed for previous events. As a result, the companion did not accrete as much material as in previous events. All this may be the consequence of just one elementary change, namely a strong decrease in the primary's mass-loss rate. This would mark the beginning of a new phase, in which the spectroscopic events can be described as an occultation by the primary's wind.